Remote Sensing of Environment: Sciencedirect

Remote Sensing of Environment 264 (2021) 112617
Contents lists available at ScienceDirect
Remote Sensing of Environment

journal homepage: www.elsevier.com/locate/rse
Air pollution scenario over Pakistan: Characterization and ranking of

extremely polluted cities using long-term concentrations of aerosols and
trace gases
Muhammad Bilal a, Alaa Mhawish a, Janet E. Nichol b, Zhongfeng Qiu a, *, Majid Nazeer c,
Md. Arfan Ali a, Gerrit de Leeuw d, e, f, g, Robert C. Levy h, Yu Wang a, Yang Chen i, Lunche Wang j,
Yuan Shi k, Max P. Bleiweiss l, Usman Mazhar m, Luqman Atique n, Song Ke o
a
Lab of Environmental Remote Sensing (LERS), School of Marine Sciences, Nanjing University of Information Science and Technology, Nanjing 210044, China
b
Department of Geography, School of Global Studies, University of Sussex, Brighton BN19RH, UK
c
Key Laboratory of Digital Land and Resources, East China University of Technology, Nanchang 330013, China
d
Royal Netherlands Meteorological Institute (KNMI), R & D Satellite Observations, 3730AE De Bilt, the Netherlands
e
School of Atmospheric Physics, Nanjing University of Information Science and Technology, Nanjing 210044, China
f
Aerospace Information Research Institute, Chinese Academy of Sciences (AirCAS), No.20 Datun Road, Chaoyang District, Beijing 100101, China
g
School of Environment Science and Spatial Informatics, University of Mining and Technology, Xuzhou, Jiangsu 221116, China
h
Climate and Radiation Laboratory, NASA Goddard Space Flight Center, Greenbelt, MD, USA.
i
State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China.
j
Laboratory of Critical Zone Evolution, School of Earth Sciences, China University of Geosciences, Wuhan 430074, China.
k
Institute of Future Cities, The Chinese University of Hong Kong, Hong Kong, SAR, China
l
Department of Entomology, Plant Pathology and Weed Science, New Mexico State University, Las Cruces, NM 88003, USA
m
School of Earth Sciences, Zhejiang University, Hangzhou 310027, China
n
School of Remote Sensing & Geomatics Engineering, Nanjing University of Information Science and Technology, Nanjing 210044, China
o
Geological Survey of Jiangsu Province, Nanjing, 210018, China
A R T I C L E I N F O A B S T R A C T
Editor: Menghua Wang Pakistan ranks third in the world in terms of mortality attributable to air pollution, with aerosol mass concen
trations (PM2.5) consistently well above WHO (World Health Organization) air quality guidelines (AQG).
Keywords: However, regulation is dependent on a sparse network of air quality monitoring stations and insufficient ground
MODIS data. This study utilizes long-term observations of aerosols and trace gases to characterize and rank the air
AOD
pollution scenarios and pollution characteristics of 80 selected cities in Pakistan. Datasets used include (1) the
CAMS
Aqua and Terra (AquaTerra) MODIS (Moderate Resolution Imaging Spectroradiometer) Level 2 Collection 6.1
MERRA-2
PM1 merged Dark Target and Deep Blue (DTB) aerosol optical depth (AOD) retrieval products; (2) the CAMS
PM2.5 (Copernicus Atmosphere Monitoring Service) reanalysis PM1, PM2.5, and PM10 data; (3) the MERRA-2 (Modern-
PM10 Era Retrospective analysis for Research and Applications, Version 2) reanalysis PM2.5 data, (4) the OMI (Ozone
OMI Monitoring Instrument) tropospheric vertical column density (TVCD) of nitrogen dioxide (NO2), and VCD of
NO2 sulfur dioxide (SO2) in the Planetary Boundary Layer (PBL), (5) the VIIRS (Visible Infrared Imaging Radiometer
SO2 Suite) Nighttime Lights data, (6) MODIS Collection 6 Version 2 global monthly fire location data (MCD14ML),
PSCF (7) population density, (8) MODIS Level 3 Collection 6 land cover types, (9) AERONET (AErosol RObotic
Pakistan
NETwork) Version 3 Level 2.0 data, and (10) ground-based PM2.5 concentrations from air quality monitoring
stations. Potential Source Contribution Function (PSCF) analyses were performed by integrating with ground-
based PM2.5 concentrations and the NOAA (National Oceanic and Atmospheric Administration) HYSPLIT
(Hybrid Single-Particle Lagrangian Integrated Trajectory) air parcel back trajectories to identify potential
pollution source areas which are responsible for extreme air pollution in Pakistan. Results show that the ranking
of the top polluted cities depends on the type of pollutant considered and the metric used. For example, Jhang,
Multan, and Vehari were characterized as the top three polluted cities in Pakistan when considering AquaTerra
DTB AOD products; for PM1, PM2.5, and PM10 Lahore, Gujranwala, and Okara were the top three; for
* Corresponding author.
E-mail address: [email protected] (Z. Qiu).
https://doi.org/10.1016/j.rse.2021.112617
Received 17 June 2021; Received in revised form 17 July 2021; Accepted 19 July 2021
Available online 27 July 2021
0034-4257/© 2021 Elsevier Inc. All rights reserved.
M. Bilal et al. Remote Sensing of Environment 264 (2021) 112617
tropospheric NO2 VCD Lahore, Rawalpindi, and Islamabad and for PBL SO2 VCD Lahore, Mirpur, and Gujran
wala. The results demonstrate that Pakistan's entire population has been exposed to high PM2.5 concentrations
for many years, with a mean annual value of 54.7 μg/m3, over all Pakistan from 2003 to 2020. This value exceeds
Pakistan's National Environmental Quality Standards (Pak-NEQS, i.e., <15 μg/m3 annual mean) for ambient air
defined by the Pakistan Environmental Protection Agency (Pak-EPA) as well as the WHO Interim Target-1 (i.e.,
mean annual PM2.5 < 35 μg/m3). The spatial analyses of the concentrations of aerosols and trace gases in terms
of population density, nighttime lights, land cover types, and fire location data, and the PSCF analysis indicate
that Pakistan's air quality is strongly affected by anthropogenic sources inside of Pakistan, with contributions
from surrounding countries. Statistically significant positive (increasing) trends in PM1, PM2.5, PM10, tropo
spheric NO2 VCD, and SO2 VCD were observed in ~89%, ~67%, ~48%, 91%, and ~ 88% of the Pakistani cities
(80 cities), respectively. This comprehensive analysis of aerosol and trace gas levels, their characteristics in
spatio-temporal domains, and their trends over Pakistan, is the first of its kind. Results will be helpful to the
Ministry of Climate Change (Government of Pakistan), Pak-EPA, SUPARCO (Pakistan Space and Upper Atmo
sphere Research Commission), policymakers, and the local research community to mitigate air pollution and its
effects on human health.
1. Introduction (2013) analyzed MODIS (Moderate Resolution Imaging Spectroradi

ometer) AOD (Aerosol Optical Depth) retrievals over Lahore and Karachi
With the rapid increase in population and overexploitation of natural from 2001 to 2010 and reported higher aerosol loadings near the city
resources, air pollution is a serious global environmental concern. Ac center than outside the city. Tariq et al. (2016) analyzed ground-based
cording to the World Health Organization (WHO, 2018a), air pollution and satellite-based aerosol optical properties over Lahore during
levels are dangerously high worldwide as 9 out of 10 people breathe intense haze events in October 2013 and reported crop residue burning
polluted air, and each year 7 million deaths are caused by outdoor and and urban-industrial emissions as the main sources of high AOD levels.
indoor aerosol pollutants. Outdoor (ambient) air pollution is due to high Bilal et al. (2016) evaluated the performance of the Aqua-MODIS
concentrations of different species including airborne particulate matter (MYD04) level 2 aerosol products over Lahore and Karachi from 2007
(PM), ozone (O3), nitrogen dioxide (NO2), volatile organic compounds to 2013, and recommended the use of Dark Target (DT) and Deep Blue
(VOC), carbon monoxide (CO), and sulfur dioxide (SO2), which have (DB) algorithms over Karachi and Lahore, respectively, for regional air
adverse health effects (Mannucci and Franchini, 2017). Although air quality applications, as these cities have different land cover charac
pollution is a global problem, the latest WHO air quality database re teristics and aerosol types. Other remote sensing studies have been
veals that 97% of affected cities are in low- and middle-income countries conducted on atmospheric trace gases, such as ozone (O3), nitrogen
with more than 100,000 inhabitants (WHO, 2018b). Air pollution is dioxide (NO2), sulfur dioxide (SO2), and carbon dioxide CO2, as well as
endemic to Pakistan, being listed among low- and middle-income their trends over time (Khokhar et al., 2016; Khokhar et al., 2015; Tariq
countries as well as being the most urbanized of its South Asian coun and Ali, 2015; ul-Haq et al., 2017; ul-Haq et al., 2014; ul-Haq et al.,
terparts (77.42 million or 36.37% of the urban population, with 2.52% 2015). Zhang et al. (2020) conducted the first study of the vertical dis
annual growth rate) (UNDP, 2019). Purohit et al. (2013) predicted that tribution of aerosol optical properties over Pakistan using CALIPSO
under current emission control standards, air pollution would decrease (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation)
life expectancy by more than 100 months by 2030. The Health Effects data.
Institute (2019) reported that since 1990, Pakistan's entire population Cities are areas of high activity, and every city is a huge source of
has been exposed to PM2.5 (the integrated dry mass of aerosol particu local anthropogenic aerosols and trace gases from industrial and human
lates with an aerodynamic diameter less than 2.5 μm) annual mean activities, which can impact air quality, visibility, and alter the physico-
concentrations of 58 μg/m3 in 2017, levels exceeding WHO Interim chemical properties of the atmosphere at local, regional, and global
Target-1 (i.e., <35 μg/m3). Pakistan ranks third in the world in terms of scales. Although several studies of AOD and atmospheric trace gases
mortality attributable to air pollution, with an annual loss of 128,000 have been conducted over Pakistan, no study has encompassed different
lives (Government of Pakistan, F.D, 2019). Recently, on October 30, particle fractions (PMx, x = 1, 2.5, and 10) on the national scale, i.e., the
2019, the Air Quality Index (AQI) was 484 in Lahore (the second-largest dry mass of ultrafine particles with an aerodynamic diameter less than 1
city with the highest urbanization rate of 6.12% per annum), well above μm (PM1), 2.5 μm (PM2.5) and 10 μm (PM10). PM1 is part of PM2.5, PM2.5
the threshold of 300 for “hazardous” level (Amnesty International, is part of PM10. It is of great importance to identify the cities most
2019). The winter of 2019–2020 witnessed a spate of smog, which affected by different PMx fractions, as they have different effects on, for
compelled authorities in Punjab to close schools for an extended period. instance, health and chemical and physical processes in the atmosphere,
The formation of this smog was fueled by the buildup of anthropogenic and this is the first study to do so. Moreover, very few studies have
aerosols having 65% of sources within Pakistan. The principal cause for investigated the long-term trends in pollutant concentrations at city
smog formation is NOx, which is emitted primarily from Pakistan's 23.6 level, which can provide additional insights into the link between con
million transport vehicles (58%), followed by industry and power, centrations and the changes in emissions. Furthermore, previous studies
which accounts for 34% of emissions (Amnesty International, 2019; are not comprehensive enough to answer questions such as: which are
Government of Pakistan, F.D, 2019; UNDP, 2019). According to the the most and least polluted cities of Pakistan, and what are the likely
Pakistan Air Quality Initiative (PAQI), Lahore, Peshawar, Islamabad, pollution sources? Therefore, this study aims (1) to extensively charac
and Karachi are the most polluted cities where air quality does not meet terize and rank the extremely polluted cities of Pakistan, considering
WHO air quality guidelines during autumn and winter (PAQI, 2018). Air multiple sources and aerosol mass fractions, for 80 carefully selected
pollution monitoring throughout Pakistan is challenging due to sparsely cities, representing almost all major urban centers of Pakistan, and (2) to
distributed air quality monitoring stations, though several remote identify the likely pollutant sources by performing PSCF (Potential
sensing studies have been conducted. Source Contribution Function) analysis with the integration of HYSPLIT
Satellite observations provide spatial distributions of column- (Hybrid Single-Particle Lagrangian Integrated Trajectory) back trajec
integrated concentrations which are related to the near-surface con tory and ground-based PM2.5 concentrations. This study is based on
centrations through meteorological and physico-chemical processes, long-term combined Aqua and Terra (AquaTerra) MODIS data from
thus complementing local ground-based observations. Gupta et al. 2003 to 2017, OMI (Ozone Monitoring Instrument) data (NO2 and SO2)
2
from 2004 to 2019, CAMS (Copernicus Atmosphere Monitoring Service) AERONET sites are located in an urban area, and approximately 20 km
reanalysis PM1, PM2.5, and PM10 data from 2003 to 2019, MERRA-2 away from the Arabian Sea coast, respectively.
(Modern-Era Retrospective analysis for Research and Applications,
Version 2) PM2.5 data from 2003 to 2020, VIIRS (Visible Infrared Im 3.2. AquaTerra MODIS data
aging Radiometer Suite) Nighttime Lights from 2012 to 2019, LandScan
global population density for 2019, MODIS land cover type for 2019, In the present study, Aqua and Terra MODIS C6.1 L2 aerosol prod
MODIS global monthly fire location data from 2003 to 2020, ground- ucts at 10 km spatial resolution are obtained from 2003 to 2017 for
based PM2.5 concentrations from 2018 to 2020, and AERONET (AEro Pakistan from the LAADS DAAC (https://ladsweb.modaps.eosdis.nasa.
sol RObotic NETwork) AOD measurements from 2006 to 2017. Detailed gov/). The MODIS aerosol product provides DT AOD retrievals over
information on the data used in this study is provided in Section 3. land and water surfaces (Levy et al., 2013), and DB AOD retrievals only
over land (Hsu et al., 2013). The DT and DB AOD retrievals for different
2. Study area collections are extensively validated against Sunphotometer (AERO
NET) measurements at regional (Bilal et al., 2019b; Bilal et al., 2014;
Pakistan, with a population of 212.82 million, is the sixth most Che et al., 2019; de Leeuw et al., 2018; Fan et al., 2017; Filonchyk et al.,
populous country in the world. It lies between 23◦ 35′ to 37◦ 05′ North 2019; Gupta et al., 2013; He et al., 2018; Islam et al., 2019; Livingston
and 60◦ 50′ to 77◦ 50′ East, having a diverse geographical landscape et al., 2014; Mhawish et al., 2017; More et al., 2013; Nichol and Bilal,
bordered by China, the Himalayas, India, Afghanistan, Iran, and the 2016; Shen et al., 2018; Shi et al., 2013; Sogacheva et al., 2018; Wang
Arabian Sea. Geographically, Pakistan falls into three major regions: the et al., 2017; Wang et al., 2019; Xiao et al., 2016; Xie et al., 2011) and
northern highlands, constituting parts of the Hindu Kush, the Karakoram global scales (Bilal et al., 2018a; Bilal et al., 2017; Levy et al., 2013; Levy
Range, and the Himalayas; the Indus River basin plain in the center and et al., 2010; Mehta et al., 2016; Remer et al., 2013; Sayer et al., 2013;
east (65% of the total area i.e. 796,096 km2); and the Balochistan Sayer et al., 2014; Sayer et al., 2015; Tong et al., 2020). These studies
Plateau in the south and west (Government of Pakistan, F.D, 2019). have reported overestimation and underestimation in DT and DB AOD
Administratively, Pakistan has six units: Punjab, Sindh, Khyber Pak retrievals respectively, due to error in the estimated surface reflectance
htunkhwa, Balochistan, Azad Kashmir, and Gilgit Baltistan. Punjab is the and aerosol scheme used in the inversion methods, but overall their
most populous (112.38 million; 53%) administrative unit of Pakistan, performance is satisfactory. Previous studies (Bilal et al., 2018a; Bilal
followed by Sindh (49.05 million; 23%), Khyber Pakhtunkhwa (36.5 and Nichol, 2017; Bilal et al., 2017; Bilal et al., 2018b; Mei et al., 2019;
million; 17%), and Balochistan (12.7 million; 6%). Balochistan has the Sayer et al., 2014) have also reported different spatial coverage of DT
largest area (43.6%), followed by Punjab (25.8%), Sindh (17.7%), and and DB AOD retrievals over land due to differences in their approaches,
Khyber Pakhtunkhwa (12.78%). Sindh is the most urbanized and i.e., pixel selection criteria, estimation of surface reflectance, and the
industrialized administrative unit of Pakistan with 52% urban popula cloud mask. Therefore, a new merged Scientific Data Set (SDS: AOD 550
tion. Islamabad (2.1 million; 1%) Capital Territory (ICT), a rather small Dark Target Deep Blue Combined) was introduced which contains only
unit in terms of area (0.1%), is, in fact, the second most urbanized the highest quality DT and DB (DTB) AOD retrievals or their average
(50.58%) region of Pakistan, and has an annual urbanization rate of values (Levy et al., 2013). The purpose of this new dataset is to improve
4.91%. Currently, 10 cities in Pakistan have a population of over one spatial coverage over land (Levy et al., 2013; Sayer et al., 2014), i.e., to
million, and 7 have higher per-capita incomes than the national average retrieve AOD in the same image for those regions where either the DT or
(UNDP, 2019). The Pakistan economic survey 2018–19 reports a total the DB algorithm does not achieve a successful retrieval (Bilal et al.,
cropped area of 22.6 million hectares, and agricultural contributions of 2017; Levy et al., 2013). The merged DTB AOD retrievals have been
18.5% to the GDP, compared with 20.3% from the industrial sector validated at regional and global scales (Ali and Assiri, 2019; Bilal et al.,
(Government of Pakistan, F.D, 2019). 2018a; Bilal and Nichol, 2017; Bilal et al., 2017; Sayer et al., 2014;
This study covers almost all prominent cities in Pakistan including all Sogacheva et al., 2018). However, the new customized method-1 (CM1)
administrative units and their capital cities, and the Capital of the (Bilal et al., 2017), which is named Simplified Merge Scheme (SMS) in
country (Fig. 1). In summary, the study area analyzes 23 cities from the the later publications (Bilal et al., 2018a; Bilal et al., 2018b), provides
most populated administrative unit, Punjab; Khyber Pakhtunkhwa is equally consistent data quality with the combined DTB AOD retrievals
also well-represented by 19 urban centers; Balochistan is the least available in C6.1, but with significantly improved spatio-temporal
populated but the largest administrative unit, and is represented by 19 coverage.
cities; 14 other cities exemplify the diversity of Sindh in the South-East,
and 5 cities represent the attractive hilly land of Azad Kashmir. 3.3. CAMS data
3. Dataset The Copernicus Atmosphere Monitoring Service (CAMS) reanalysis is

an atmospheric composition dataset generated by the European Centre
3.1. AERONET data for Medium-Range Weather Forecasts (ECMWF). The global CAMS
model combines satellite-based observations with chemistry-aerosol
The AERONET (AErosol RObotic NETwork) (Holben et al., 1998; modeling using the four-dimensional variational (4D-VAR) data assim
Holben et al., 2001) is a global network of calibrated Sunphotometers ilation technique to obtain the mass concentration of aerosols and trace
coordinated by NASA (National Aeronautics and Space Administration) gases. CAMS uses the MACCity inventory at 0.5◦ × 0.5◦ spatial resolu
which provides regular measurements of spectral AOD at 340 nm, 380 tion for anthropogenic emissions which covers the period 1960–2010
nm, 440 nm, 500 nm, 675 nm, 870 nm, 1020 nm, and 1640 nm, and AE (Granier et al., 2011). Detailed information about the model and the
at 340–440 nm, 380–500 nm, 440–675 nm, and 500–870 nm at three emission inventory can be found in (Flemming et al., 2017; Flemming
levels, i.e., Level 1.0 (unscreened), Level 1.5 (cloud-screened), and Level et al., 2015). In this study, the ground-based mass concentration of
2.0 (cloud-screened and quality-assured), under cloud-free skies (Smir particulate matter, including particles with an aerodynamic diameter of
nov et al., 2000) for every 15 min with an uncertainty of 0.01–0.02 less than 1 μm (PM1), less than 2.5 μm (PM2.5), and less than 10 μm
(Holben et al., 2001). The present study used Version 3 Level 2.0 AOD at (PM10) was obtained from the CAMS reanalysis data for the years 2003
500 nm (AOD500) and AE at 440–675 nm (AE440–675) (Giles et al., 2019) and 2020. PMx (x = 1, 2.5, & 10) data were used at two different
obtained from the AERONET website (https://aeronet.gsfc.nasa.gov/) spatiotemporal resolutions, i.e., (i) CAMS global reanalysis dataset at
for the Lahore (31.47987ο N, 74.26406ο E) and Karachi (24.94574ο N, 0.75◦ × 0.75◦ spatial resolution and 3-hourly temporal resolution from
67.13594ο E) sites from 2006 to 2017. The Lahore and Karachi 2003 to 2020, and (ii) CAMS near-real time dataset at 0.125◦ × 0.125◦
3
(caption on next page)
4
Fig. 1. Geographical and administrative map of Pakistan including a list of cities used in the present study. Cities are characterized using (a) yearly mean CAMS
(Copernicus Atmosphere Monitoring Service) reanalysis PM2.5 concentrations (μg/m3) for the years 2003 and 2020, and (b) yearly mean AquaTerra MODIS DTB AOD
retrievals at 550 nm from 2003 to 2017. Extremely polluted cities (red colour) are defined for PM2.5 > 92.84 (AOD > 0.6) (3rd quartile), highly polluted cities (brown
colour) for 45.69 ≤ PM2.5 ≤ 92.84 (0.3 < AOD < 0.6) (between 3rd and 1st quartiles), and polluted cities (purple colour) for PM2.5 < 45.69 (AOD < 0.3) (1st quartile)
using descriptive statistics (Table S1). Cities are not defined as low polluted or clean cities as annual mean PM2.5 concentrations for all cities exceed Pakistan's
National Environmental Quality Standards (Pak-NEQS) for ambient air (<15 μg/m3 annual mean). (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)
spatial resolution and 12-hourly temporal resolution from 2018 to 2020 spectral regions to aerosol absorption, and it retrieves absorbing aerosol
(Inness et al., 2019). The PMx data at 0.75◦ grid size and 3-hourly optical depth (AAOD) at 388 nm (Torres et al., 2013; Torres et al., 2007).
temporal resolution were used for long-term climatology and for char Along with the AAOD, the OMAERUV algorithm also provides an ul
acterizing extremely polluted cities, whereas, the CAMS near-real time traviolet Aerosol Index (UVAI), AOD, and Single Scattering Albedo
data at 0.125◦ grid size and 12-hourly temporal resolution were used for (SSA). OMI also retrieves the atmospheric trace gases O3, NO2 and SO2
validation against ground-based PM2.5 concentrations obtained from air (Carn et al., 2017; Krotkov et al., 2017; Krotkov et al., 2016; Li et al.,
quality monitoring stations. 2017; Li et al., 2013; Veefkind et al., 2006). In this study, OMAERUV
version 3 Level 3 daily cloud-screened (cloud fraction <30%) NO2
3.4. MERRA-2 reanalysis data tropospheric vertical column density (TVCD) (OMNO2e), and SO2 VCD
in the planetary boundary layer (PBL) (OMSO2e) gridded at 0.25◦ ×
The MERRA-2 (Modern-Era Retrospective analysis for Research and 0.25◦ spatial resolution from 2004 to 2019 were used.
Applications, Version 2) atmospheric reanalysis is the latest data
released by the NASA GMAO (Global Modeling and Assimilation Office) 3.7. Other supporting datasets
in 2017 (Buchard et al., 2017; Randles et al., 2017). The MERRA-2
aerosol gridded data, i.e., dust, sea salt, sulfate, black carbon, and Other supporting datasets include (i) annual mean VIIRS nighttime
organic carbon, are simulated with 72 vertical layers from the surface to lights data (https://eogdata.mines.edu/products/vnl/) from 2012 to
higher than 80 km using the GEOS-5 (GMAO Earth system model version 2019 derived from monthly mean data (Elvidge et al., 2021), (ii) MODIS
5) model radiatively coupled to the GOCART (Goddard Chemistry Collection 6 global monthly Fire Location product (MCD14ML) from
Aerosol Radiation and Transport) model (Chin et al., 2002; Colarco 2003 to 2020 (https://firms.modaps.eosdis.nasa.gov/download/), (iv)
et al., 2010). For anthropogenic emissions, MERRA-2 uses the EDGAR- MODIS Collection 6 Level 3 land cover type product (MCD12Q1) for
4.2 emission inventory at 0.1◦ × 0.1◦ spatial resolution which covers 2019 (https://ladsweb.modaps.eosdis.nasa.gov/), and (v) the LandScan
the period 1970–2008 (Janssens-Maenhout et al., 2013). In this study, population density (https://landscan.ornl.gov/) for 2019 (Rose et al.,
the MERRA-2 aerosol gridded data (dust, sea salt, sulfate, black carbon, 2020).
and organic carbon) at 0. 5◦ × 0.625◦ spatial resolution from 2018 to
2020 were used. More details about MERRA-2 reanalysis data can be 4. Research methodology
found in Randles et al. (2017) and Buchard et al. (2017).
To investigate the air pollution scenario over Pakistan and charac
3.5. Ground-based PM2.5 measurements terize the extremely polluted cities, in this study the following meth
odology was adopted:
Ground-based PM2.5 measurements were obtained from two
different air quality monitoring networks. Firstly, PM2.5 data were ob 1. MODIS AOD retrievals were obtained from the Scientific Data Set
tained from 4 air quality stations operated by the US Consulates in (SDS) “Optical Depth Land and Ocean” and “Deep Blue Aerosol
Islamabad, Karachi, Lahore, and Peshawar, and secondly, 54 air quality Optical Depth 550 Land Best Estimate”. Only the highest quality-
monitoring stations operated by PAQI in Lahore (24 stations), Karachi assured DT (QA = 3) and DB (QA ≥ 2) retrievals were used, as rec
(15), Islamabad (5) Sialkot (3), Peshawar (2), Rawalpindi (2), Faisala ommended by previous studies (Bilal et al., 2013; Levy et al., 2013;
bad (1), Gujranwala (1), and Muridke (1). Due to the lack of a well- Mhawish et al., 2019; Sayer et al., 2013). Pakistan has a variety of
developed and standard air quality network of ground-based PM2.5 land cover types, e.g., snow and mountainous land surface in
measurements, this study is limited to only these cities for the validation Northern Pakistan, plain and agricultural land surfaces in Central
of CAMS and MERRA-2 reanalysis PM2.5 gridded data. PM2.5 concen Pakistan, and arid and desert land surfaces in southern Pakistan,
trations from the US Consulates are measured by beta gauge attenuation where the DT and DB algorithms overestimate and underestimate,
monitors (BAM-1020; Met One Instruments), hereafter referred to as respectively. However, the DT algorithm is unable to provide re
BAM PM2.5 concentrations. To increase social awareness in Pakistan, trievals over the arid and desert land surfaces of Balochistan. Similar
PAQI provides PM2.5 data using a nationwide network of low-cost air results were observed and reported in our previous study over
quality monitors (IQAir AirVisual Pro), hereafter referred to as LCM Pakistan (Bilal et al., 2016). Therefore, in the present study, we
PM2.5 concentrations. In this study, LCM and BAM PM2.5 measurements preferred to generate the combined (merged) DTB AOD550 retrievals
were used for January 2018–December 2019 and January for both Aqua and Terra MODIS data from 2003 to 2017 using the
2019–February 2021, respectively. More details about PAQI (LCM) and customized method-1 (CM1) (Bilal et al., 2017), which in later
US Consulates (BAM) PM2.5 data can be found in Shi et al. (2020) and publications is named Simplified Merge Scheme (SMS) (Bilal et al.,
Mhawish et al. (2020), respectively. 2018a; Bilal et al., 2018b), i.e., an average of the DT and DB AOD
retrievals or the available one with the highest quality flag (Eq. 1), to
3.6. OMI data enhance spatio-temporal coverage.
⎧ ⎫
if only DT AOD exists→DT
The Ozone Monitoring Instrument (OMI) onboard the Aura satellite ⎨ ⎬
DTB AOD550 = if only DB AOD exists→DB (1)
was launched in July 2004 as a part of the A-Train satellite constellation. ⎩ ⎭
if both DT and DB AOD exist→(DT + DB)/2
OMI is a hyperspectral sensor that measures the radiation reflected from
the earth-atmosphere system, in the wavelength range 250–500 nm and
provides daily global coverage at a spatial resolution of 13 × 24 km2 at 2. Aqua and Terra MODIS may not provide complete spatial coverage
nadir. The OMI OMAERUV algorithm utilizes the sensitivity of near-UV due to cloud cover. On days when Aqua provides AOD retrievals,
5
Terra may not, and vice-versa. Therefore, for more complete spatial estimations.
coverage between Aqua and Terra as well as to represent an average σMODISAOD
air pollution scenario between morning and afternoon times with a β= (6)
σ AERONET AOD
single dataset, the combined AquaTerra DTB AOD retrievals were
generated from the Aqua DTB and Terra DTB AOD retrievals using (
σMODISAOD
)
SMS/CM1, i.e., an average of the Aqua and Terra DTB AOD retrievals α = MODISAOD − × AERONET AOD (7)
σ AERONET AOD
or the available one (Eq. 2).
⎧ ⎫ Where, β, α, σ MODISAOD, and σ AERONETAOD are the slope, intercept, the
⎨ if only Aqua AOD exists→Aqua ⎬ standard deviation of MODIS AOD, and standard deviation of AERONET
AquaTerra AOD =
⎩
if only Terra AOD exists→Terra
⎭ AOD, respectively.
if both Aqua and Terra AOD exist→(Aqua + Terra)/2
(2) 5. To show the long-term variation of the mean spatial distributions of
AquaTerra AOD over Pakistan, the AOD retrievals from 2003 to 2017
3. The AquaTerra DTB AOD retrievals are validated against Sunpho are used to generate monthly mean spatial AOD maps, and their
tometer AOD measurements obtained for Lahore (31.480◦ N and corresponding pixel counts are calculated for reporting the retrieval
74.264◦ E) and Karachi (24.946◦ N and 67.136◦ E) AERONET sites. performance of both the DT and DB algorithms.
The AERONET Sunphotometer does not provide AOD at 550 nm 6. To assure the quality of the PM2.5 data, validation of daily average
(AOD550), AOD550 is interpolated using AOD at 500 nm (AOD500) CAMS and MERRA-2 PM2.5 data was conducted against in-situ PM2.5
and Ångström Exponent at 440–675 nm (AE440–675) based on the measurements obtained from the air quality monitoring stations. The
Ångström Exponent empirical formula (Eq. 3) (Eck et al., 1999). performance was evaluated based on the correlation coefficient (r),
Collocated AquaTerra and AERONET AOD retrievals were defined as RMB (Eq. 5), and slope (Eq. 6). MERRA-2 PM2.5 concentrations were
the average of at least two pixels of DTB within a spatial region of 3 calculated based on five aerosol components using Eq. 8 (Song et al.,
× 3 pixels (at least 2 out of 9 pixels) centered on the AERONET site 2018), and CAMS PM2.5 and PM10 concentrations were calculated
and the average of at least two AERONET AOD measurements be using Eqs. 9 and 10 (Rémy et al., 2019).
tween 10:00 and 14:30 local solar time. PM 2.5 = [Dust2.5 ] + [SS2.5 ] + 1.375 × [SO4 ] + [BC] + 1.6 × [OC] (8)
( )− AE440− 667
AOD550 = AOD500
550
(3) Where, Dust2.5, SS2.5, BC, OC, and SO4 are the GOCART concentra
500 tions of dust, sea salt, black carbon, organic carbon, and sulfate in par
ticles with a diameter smaller than 2.5 μm, respectively.
4. Accuracy and errors are reported using the Pearson correlation co PM 2.5 = ρ([SS1 ]/4.3 + [SS2 ]/4.3 + [DD1 ] + [DD2 ] + 0.7[OM] + [BC]
efficient (r), the expected error (EE, Eq. 4), and relative mean bias + 0.7[SU] + 0.7[NI 1 ] + 0.25[NI 2 ] + 0.7[AM] ) (9)
(RMB, Eq. 5). The slope (β, Eq. 6) and intercept (α, Eq. 7) between
collocated AquaTerra DTB and AERONET AOD data are calculated
using the reduced major axis (RMA) regression which incorporates
errors in both independent (AERONET) and dependent (MODIS) PM 10 = ρ([SS1 ]/4.3 +[SS2 ]/4.3 +[DD1 ]+[DD2 ]+0.4[DD3 ]+[OM]+[BC]
variables (Bilal et al., 2019a; Harper, 2016). The performance of the +[SU]+[NI 1 ]+ [NI 2 ]+[AM]) (10)
Terra and Aqua DT, DB, and DTB AOD retrievals is evaluated based Where [SS1,2] = sea salt aerosol, [DD1,2,3] = desert dust, [NI1,2] =
on (i) highest correlation coefficient (r), (ii) highest number of nitrate, [OM] = organic matter, [BC] = black carbon, [SU] = sulfate,
collocated retrievals (N), (iii) the highest percentage of retrievals and [AM] = ammonium (concentrations in particles with a diameter
within the EE, and (iv) lowest RMB. To evaluate the performance of smaller than 2.5 μm from the CAMS model).
the collocated retrievals, the following criteria are utilized (Bilal
et al., 2017): the DT, DB, and DTB retrievals are considered to be of 7. To characterize extremely polluted cities in Pakistan, the DTB AOD
equal quality if the relative difference is within (1) 5% for the cor retrieved from AquaTerra, the PM1, PM2.5, and PM10 from CAMS
relation coefficient (r), (2) 10% for the collocated retrievals, (3) 10% data, and the SO2 VCD and NO2 TVCD from OMI are used. Polluted
for the percentage of retrievals is within the EE, and (4) RMB < 25%. months as well as years, for the corresponding polluted cities, are
EE = ± (0.05 + 0.20 × AERONET AOD ) (4) also characterized based on each pollutant.
8. To assess recent changes in the concentrations of atmospheric con
The upper and lower EE envelopes are calculated using Eqs. 4a and stituents, the non-parametric Mann Kendal test (Kendall and Gib
4b. bons, 1990; Mann, 1945) associated with Theil-Sen's slope (Sen,
Upper EE envelope = AERONET AOD + |EE| (4a) 1968; Theil, 1992) was used to estimate and detect trends over the
main cities of Pakistan from 2003 to 2020. The non-parametric Mann
Lower EE envelope = AERONET AOD − |EE| (4b)
Kendal test is often used to detect monotonic trends in a time series
The percentage of best retrieved MODIS AOD retrievals within the EE and is also suitable for non-normally distributed data, or if the data
is reported using Eq. 4c. have some missing observations such as environmental data.
Further, the bootstrapping technique was used to eliminate serial
%EE = AERONET AOD − |EE| ≤ MODISAOD ≤ AERONET AOD + |EE| (4c)
autocorrelation in the monthly mean aggregated time series data and
Where |EE| is the absolute value of EE. increase the robustness of the test (Hamed and Ramachandra Rao,
( ) 1998; Salmi et al., 2002). The significance of the calculated trend
MODISAOD − AERONET AOD was assessed using the two-tailed test method at a 95% confidence
RMB = × 100 (5)
AERONET AOD interval.
9. The NOAA (National Oceanic and Atmospheric Administration)
Where, MODISAOD and AERONET AOD are the mean of MODIS and HYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory
AERONET AOD retrievals, respectively. RMB > 0 represents over Model) (Stein et al., 2015), a complete transport, dispersion, and
estimation in MODIS AOD compared to AERONET AOD, RMB < 0 rep chemical transformation model, is used for back trajectory analysis
resents underestimation, and RMB = 0 represents no over- and under- to determine the origin of air masses (Fleming et al., 2012) and
6
highlight the possible sources of aerosol pollutants affecting the air averaged out in the DTB AOD product, as the overestimations and un
quality of Pakistan using the PSCF (Potential Source Contribution derestimations are fewer than for DT and DB, individually. Furthermore,
Function) analysis. In this study, 72 h HYSPLIT backward trajectories the RMB (− 0.03%) is significantly improved, being 99.9% and 99.8%
at the height of 500 m above the ground level (AGL) were computed lower than for DT and DB, respectively. These results indicate the better
for every 6 h at seasonal scales from March 2020 to February 2021 performance of the Terra DTB AOD product as compared to DT and DB
using the GDAS (Global Data Assimilation System) meteorological over Pakistan. Similar to Terra, the performance of the Aqua DTB AOD
data at 1◦ × 1◦ spatial resolution (available at ftp://arlftp.arlhq.noaa product (Fig. 2f) is much better than for DT (Fig. 2d) and DB (Fig. 2e)
.gov/pub/archives/gdas1). The PSCF analysis was performed for 4 products, with a significantly higher number of collocated AOD values
cities selected because of the availability of ground-based PM2.5 and lower RMB. However, Aqua performs equally as Terra in terms of
measurements from the air quality stations operated by the US correlation and the percentage of retrievals within the EE. It is important
Consulates, namely, Peshawar, Islamabad, Lahore, and Karachi. The to mention that a larger number of both DT and DB AOD retrieval
height of 500 m AGL has been reported very useful as it is the products was available for Lahore than for Karachi and also that DB
approximate height of the mixing layer (Begum et al., 2005). The provides a greater number of AOD retrievals over Pakistan than DT.
backward trajectory clustering and investigation of the origins of the Based on the superior performance of the Aqua and Terra DTB AOD
particulate matter at the receptor locations were studied using retrievals, the merged AquaTerra DTB AOD product was generated for
MeteoInfo TrajStat software (Version 2.0, available at http://meteo further analysis (see Fig. S1 in the supplementary data for the validation
think.org/products/trajstat.html) (Wang et al., 2009) in conjunc of AquaTerra DTB AOD retrievals).
tion with HYSPLIT and Geographic Information System (GIS).
The PSCF analysis was performed using 24-h average ground- 5.1.2. Spatial distribution of AOD retrievals
based PM2.5 concentrations over a grid with a resolution of 0.5◦ , Fig. 3 shows the spatial distributions of the monthly mean AquaTerra
for the days that exceeded the Pak-NEQS 24-h air quality standards DTB AOD over Pakistan together with the corresponding pixel counts
(35 μg/m3). The PSCF value for a specific grid cell was calculated on (PC) averaged over the years 2003–2017. Significant monthly variations
the assumption that the trajectory endpoint is located within a cell (i, in both AOD and PC are observed. AOD retrievals are missing over the
j) and the trajectory is assumed to collect pollutants emitted from Gilgit-Baltistan and Jammu & Kashmir (disputed territory) throughout
different pocket emission sources within that cell (i, j). The PSCF the year, except for January, as the DT and DB algorithms do not provide
value can be interpreted as a conditional probability describing the AOD retrievals over high mountain regions and snow-covered surfaces.
potential contributions of a grid cell to the high PM2.5 loadings at the The presence of AOD retrievals during January is because the DB algo
receptor site. The error associated with the trajectory is proportional rithm does not use the MODIS snow mask product directly, and the in
to the distance from the receptor location (Begum et al., 2005). The ternal snow/cloud mask does not work well over these regions.
PSCF value for the ijth grid cell can be computed using Eq. 11: Surprisingly, high AOD values >1.0 are observed during June and July
/ over the Northwestern region of Khyber Pakhtunkhwa, which is a high
PSCF(i, j) = mij nij (11)
mountainous region with permanent snow cover. These high AOD
Where, nij represents the number of endpoints that fall or pass values over snow-covered regions could be due to an error in the internal
through the ijth cell and mij denotes for the number of endpoints in the ijth snow/cloud mask of the DB algorithm which has missed these pixels
cell having a higher pollutant concentration than the 24-h Pak-NEQS. during preprocessing; DT does discard bright pixels during preprocess
The uncertainty arising due to small nij is reduced by multiplying an ing. AOD >1.0 is observed in July followed by June and August over
arbitrary weight function Wi, j, which is multiplied into the PSCF. In this Punjab and Sindh, mainly attributed to hygroscopic growth of the
case, the weight function is given in Eq. (12): aerosol particles during summer relative humidity is high, similar to
⎧ other reports using MODIS and MISR aerosol products (Mehta et al.,
⎪
⎪
⎨
if nij > 3n →1.00 2016; Mhawish et al., 2021). Most of the major cities of Punjab and
if 1.5n < nij ≤ 3n →0.70 Sindh are surrounded by cropland, and the results show that high AOD
Wi,j = (12)
⎪ if n < nij ≤ 1.5n →0.42
⎪
⎩ over Pakistan follows the same spatial pattern as that of the cropland.
if nij ≤ n →0.15
The AOD over cropland is significantly higher than over non-
Where n denotes the average number of endpoints per cell, which is agricultural (i.e., mainly desert) regions throughout the year, even
calculated for each cell that has at least one endpoint. Therefore, the during late spring and summer when dust storms are considered a major
Weighted PSCF is expressed as Eq. (13): source of aerosols over Punjab and Sindh. Local production of anthro
pogenic aerosols from urban and industrial emissions and agricultural
WPSCF = Wi,j × PSCF (i, j) (13)
pre- and post-harvest burning may be responsible for the high pollution
levels over the region. Over Balochistan, especially over the desert areas,
5. Results and discussion the AOD is low compared to that in Punjab and Sindh, but still higher
than over other administrative units. Over Punjab, the highest AOD
5.1. Aqua and Terra MODIS AOD data values are observed during the post-harvest seasons, i.e., throughout
September to November, peaking in November, probably due to biomass
5.1.1. Validation of AOD products against AERONET (crop residue) burning activities (Jethva et al., 2019; Mhawish et al.,
The MODIS AOD data used in this paper were evaluated by com 2021). However, if the high AOD levels would only be due to locally
parison with the AERONET AOD data over Lahore and Karachi. The produced aerosols, the spatial patterns during each month should be
scatterplots in Fig. 2 show that Terra DT (Fig. 2a), DB (Fig. 2b), and DTB similar, but they are not. Therefore, the transboundary transport of
(Fig. 2c) retrieved AOD are equally correlated (r = 0.83) with aerosols may contribute to Pakistan's deteriorating air quality. This is
AERONET-derived AOD, and have the same percentage of retrievals confirmed by the well-known smog episodes, occurring every year over
within the EE. However, the number of collocated observations for DTB Punjab due to both local production of aerosols from crop residue
(N = 2796) is significantly higher than for DT (N = 1437) and DB (N = burning and across the border, during which atmospheric visibility is
2486) i.e., 94.6% and 12.5% more data are available from DTB than reduced to a few meters in both urban and rural areas. Overall, much
from DT and DB, respectively. The AOD retrieved from DT is signifi higher AOD levels were observed in Pakistan during June, July, and
cantly overestimated (RMB = 17.17%), with 34.38% of the data are August (summer), followed by September, October, and November
above the EE (+EE). DB underestimates the AOD (RMB = − 9.87%) with (autumn), March, April, and May (spring), and December, January, and
28.24% of the data below the EE (− EE). These uncertainties appear to be February (winter). The higher AOD in the summer is attributed to
7
Fig. 2. Validation of Terra and Aqua DT, DB, and DTB AOD products versus AERONET Version 3 Level 2.0 AOD measured in Lahore (for location, see no. 1 in Fig. 1a)
and Karachi (for location, see no. 56 in Fig. 1a) from 2006 to 2017. The red line represents the regression line, the solid black line represents the identity line, and the
dashed black lines represent the upper and lower EE envelopes. The orange points represent AOD pairs at Karachi, the blue dots at Lahore. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)
8
Fig. 3. Monthly mean spatial distributions of AquaTerra DTB AOD550 and the total number of corresponding Pixel Counts (PC) over Pakistan, both averaged over the
years from 2003 to 2017. The six units in Pakistan are indicated in the upper-left figure: GB = Gilgit-Baltistan, AK = Azad Kashmir, KP = Khyber Pakhtunkhwa, PJ =
Punjab, BL = Balochistan, and SN = Sindh.
9
several reasons, including (i) hygroscopic growth of aerosol particles, large number of AOD retrievals (>400) per pixel are available over
due to high relative humidity, which increases the extinction efficiency Balochistan and some parts of Punjab, and from late autumn to early
of the atmospheric aerosols (Dickerson et al., 1997; Li and Wang, 2014), spring, a large number of AOD retrievals (>400) per pixel are available
(ii) the enhancement of secondary aerosol formation rate due to faster over Sindh and some parts of Punjab. This could be attributed to the
photochemical reactions during higher temperatures (Jacob and seasonality in the surface albedo due to changes in vegetation cover
Winner, 2009; Kulmala et al., 2020), and (iii) the larger contribution of and/or the presence of cloud cover. Only October provides favorable
natural aerosols (mainly dust) during the summer monsoon (Mhawish conditions to both the DT and DB algorithms, when more than 400 AOD
et al., 2021). retrievals are available over Pakistan from both algorithms, except for
Fig. 3 shows a distinct pattern of PC which suggests that the DT and Gilgit-Baltistan and disputed areas, due to high surface albedo for snow/
DB algorithms do not perform equally temporally or spatially. For ice surfaces.
example, between 2003 and 2017, from late spring to early autumn, a
Fig. 4. Characterization of extremely polluted to polluted cities in Pakistan using AquaTerra DTB AOD550 products from 2003 to 2017. (a) polluted cities based on
mean AOD, (b) pixel counts, (c) polluted months based on mean AOD, and (d) polluted years based on mean AOD.
10
5.1.3. Characterization of extremely polluted cities using MODIS data with LCM (low-cost monitor) PM2.5 concentrations for 2018–2019 pro
Fig. 4a shows the mean AOD550 retrievals for 80 cities (Fig. 1) ob vided by PAQI. The scatterplots in Fig. 5 show a significant underesti
tained from the annual mean AquaTerra DTB AOD550 images and cat mation of both daily (Fig. 5a and b) and monthly (Fig. 5e and f) MERRA-
egorizes the extremely polluted to polluted cities. The thresholds for 2 PM2.5 concentrations compared to both BAM and LCM PM2.5 mea
polluted and extremely polluted cities are defined based on the values of surements: for the daily data the slopes are 0.45 and 0.54 and the RMB
first (Q1) and third (Q3) quartiles respectively, and these quartiles are are − 34.2% and − 26.8%, respectively, and for the monthly data the
calculated by analyzing descriptive statistics (Table S1) for the AOD slopes are 0.30 and 0.52 with − 35.9% to 25.3%, respectively. The re
values extracted for 80 cities. Highly polluted cities are defined based on sults also show the weak correlation of MERRA-2 PM2.5 data with both
the AOD range between the first and third quartiles. For example, AOD BAM and LCM daily (r = 0.22 and 0.25, respectively) and monthly (r =
< 0.3 (1st quartile) represents polluted cities, 0.3 ≤ AOD ≤ 0.6 (between 0.10 and 0.27, respectively) PM2.5 data. The weak correlation suggests
1st and 3rd quartiles) represents highly polluted cities and AOD > 0.6 that MERRA-2 PM2.5 data based on the GOCART aerosol module is un
represents extremely polluted cities (3rd quartile). A total of 21 cities fall able to accurately reproduce the temporal variations in PM2.5. A sig
within the category of extremely polluted cities (Punjab: 12, Sindh: 7, nificant underestimation of MERRA-2 PM2.5 data was also reported over
and Balochistan: 2), 35 cities in the category of moderately polluted China (He et al., 2019; Song et al., 2018) and India (Navinya et al.,
cities (Punjab: 11, Sindh 7, Balochistan: 7, Khyber Pakhtunkhwa: 8, 2020), but over Pakistan, the correlation is even weaker. Moreover, the
Azad Kashmir: 2), and 24 cities in the category of low polluted cities grid size of MERRA2 (0. 5◦ × 0.625◦ grid size) could introduce errors
(Punjab: 0, Sindh 0, Balochistan: 10, Khyber Pakhtunkhwa: 11, Azad due to heterogeneity within the large area that affects the correlation
Kashmir: 3). The top 3 polluted cities are Jhang, Multan, and Vehari in with the in-situ measurements.
Punjab, as Punjab is the most urbanized and populated administrative In comparison with the MERRA-2 data, the correlation coefficients of
unit (Figs. 1b and 4a), with more vehicles and industries, and also faces the CAMS daily (Fig. 5c and d) and monthly (Fig. 5g and h) PM2.5 data
severe smog episodes and dust storms, resulting in extremely high AOD versus ground-based in situ PM2.5 measurements are substantially
levels over the region. Along with anthropogenic aerosols produced higher for both BAM and LCM. However, the data in Fig. 5 show sig
locally from cropland, urban and industrial emissions, regional transport nificant deviations of the CAMS-estimated PM2.5 from the ground-based
of aerosols may be responsible for Punjab's severe air pollution problems PM2.5 values, with over- or under-estimation depending on grid size. For
which will be investigated using the PSCF analysis based on the HYSPLIT example, CAMS overestimates PM2.5 at the 0.75◦ grid size by 30.4% in
air parcel back trajectory analysis and BAM PM2.5 concentrations (see comparison with the daily BAM data and by 55.4% in comparison with
section 5.7). the daily LCM data. For monthly data, these percentages are 30.4% and
Fig. 4b shows the pixel counts (PC) of the daily AOD retrievals for 57.4%. In contrast, CAMS underestimates PM2.5 at the 0.125◦ grid size in
each city from 2003 to 2017. Results show a large number of PC for most comparison with BAM data and overestimates in comparison with LCM
cities, indicating that the characterization of extremely polluted to data. These results suggest that grid size and ground-based PM2.5 mea
polluted cities is based on a large number of PC, which supports the surement methods (BAM and LCM) play an important role in the over
results in Fig. 4a and provides confidence in the use of merged Aqua estimation/underestimation of CAMS PM2.5 data. For illustration, in
Terra DTB AOD products for quantitative research applications over comparison with the BAM PM2.5 measurements, CAMS data are over
Pakistan. However, it is noted that the lowest number of PC is observed estimated for one grid (0.75◦ ) and underestimated for another grid
for the coastal (Ormara and Gwadar) and mountainous (Dir) cities, (0.125◦ ), and CAMS PM2.5 data at the same grid size (0.125◦ ) are
where the inversion scheme of both the DT and DB algorithms needs to underestimated when compared with data measured using the BAM
be improved. method and overestimated when compared with data measured using
The monthly mean AOD retrievals are plotted to identify the high the LCM method. It is worth mentioning that both MERRA-2 and CAMS
and low polluted months in Pakistan (Fig. 4c). The months of June, July, simulate 5 types of fine particulate matter components (dust, sea salt,
and August are by far the most polluted, with AOD > 1.20 for extremely sulfate, organic carbon, and black carbon), but nitrate concentrations
polluted cities. A similar pattern of monthly variation in AOD is are not included. If the lack of nitrate concentrations is the main reason
observed for all other cities, though at lower pollution levels. As for underestimation in MERRA PM2.5 data, as reported by previous
mentioned in section 5.1.2, these months may be affected by aerosol studies (Buchard et al., 2016; He et al., 2019; Provencal et al., 2017;
pollutants from local sources such as agricultural land, urban and in Song et al., 2018), then underestimation should be observed in CAMS
dustrial regions, and deserts. Fig. 4d, showing inter-annual variations, PM2.5 data at 0.75◦ grid size, but this is not the case. Therefore, the exact
indicates very high AOD levels for extremely polluted cities throughout reasons for underestimation in both MERRA-2 and CAMS as well as
the last two decades, with annual mean AOD > 0.60, and with the most overestimation in CAMS data should be thoroughly investigated in
polluted years being 2004, 2006, 2008, 2016, and 2017. future studies. The results show a higher correlation for CAMS monthly
data (Fig. 5g and h) compared to the daily data (Fig. 5c and d). Although
5.2. CAMS and MERRA-2 reanalysis data CAMS monthly data at 0.75◦ grid size show overestimation, they have a
good correlation coefficient (r = 0.72–0.76) with ground-based PM2.5
5.2.1. Validation of PM2.5 reanalysis data measurements and could be useful for characterizing pollution levels in
Previous studies have evaluated the uncertainties in both CAMS and the cities of Pakistan compared to the MERRA-2. The comparisons in
MERRA-2 PM2.5 reanalysis data compared to ground-based PM2.5 mea Fig. 5 do not provide a strong reason for choosing one data set over the
surements (Cuevas et al., 2015; He et al., 2019; Song et al., 2018; Ukhov other. We have selected the CAMS data at the 0.75o grid taking into
et al., 2020). Recently, Ukhov et al. (2020) reported overestimation in account the deviation in the CAMS data observed in this evaluation, in
CAMS PM2.5 over the middle east and west Asia which have been addition to the large scatter in individual data points which adds
attributed to the deficient size distribution of the emitted dust. Addi uncertainty.
tionally, significant underestimation in MERRA-2 PM2.5 was reported
over China and India (He et al., 2019; Navinya et al., 2020; Song et al., 5.2.2. Characterization of extremely polluted cities using PM1 and PM2.5
2018) which could be due to the lack of nitrate concentrations in the concentrations
reanalysis data and underestimation of OC emission for urban/suburban PM1 and PM2.5 are fine particulate matter associated with human
areas (Buchard et al., 2016; Provencal et al., 2017). health issues. PM1 is more harmful than PM2.5 as it can reach deeper into
The MERRA-2 and CAMS PM2.5 reanalysis data over Pakistan were the lungs and affect the respiratory system (Liu et al., 2013; Meng et al.,
evaluated by comparison with BAM (beta gauge attenuation monitor) 2013). A previous study over China reported that most health issues
PM2.5 concentrations for 2019–2020 provided by the US Consulates and associated with PM2.5 were mainly due to greater contributions of PM1
11
Fig. 5. Validation of MERRA-2 and CAMS PM2.5 reanalysis data against BAM (beta gauge attenuation monitor) PM2.5 concentrations for 2019–2020 provided by
the US Consulates and LCM (low-cost monitor) PM2.5 concentrations for 2018–2019 provided by PAQI. Where, (a) MERRA-2 daily PM2.5 vs. BAM daily PM2.5, (b)
MERRA-2 daily PM2.5 vs. LCM daily PM2.5, (c) CAMS daily PM2.5 vs. BAM daily PM2.5, (d) CAMS daily PM2.5 vs. LCM daily PM2.5, (e) MERRA-2 monthly PM2.5 vs.
BAM monthly PM2.5, (f) MERRA-2 monthly PM2.5 vs. LCM monthly PM2.5, (g) CMAS monthly PM2.5 vs. BAM monthly PM2.5, and (h) CAMS monthly PM2.5 vs. LCM
monthly PM2.5. The dashed line in each figure is the identity line and the blue and orange solid lines are the fit lines with parameters presented in the legends. (For
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
in PM2.5 (Chen et al., 2017). The ranking of extremely polluted to diseases by 8%, 6%, and 4%, respectively, due to long-term exposure to
polluted cities in Pakistan according to annual mean CAMS PM1 con fine particulates (Pope et al., 2002).
centrations from 2003 to 2020 in Fig. 6a indicates that the top 10 Figs. 6b and 7b show months with the highest levels of PM1 and
extremely polluted cities are Lahore (135.44 μg/m3), Gujranwala PM2.5, averaged over the years 2003–2020, for the extremely polluted
(131.99 μg/m3), Okara (107.72 μg/m3), Faisalabad (98.96 μg/m3), cities. The higher PM1 and PM2.5 concentrations were observed in cold
Pakpattan (94.06 μg/m3), Jhelum (85.51 μg/m3), Sargodha (84.30 μg/ months (October to February) with the maximum concentrations in
m3), Bhimber (83.99 μg/m3), Gujrat (83.99 μg/m3), and Sialkot (83.99 December and January, while warmer months (March to September)
μg/m3). Similarly, the top 10 extremely polluted cities (Fig. 7a) ranked showed lower PMx concentrations. The high levels of fine particulates in
according to PM2.5 concentrations are Lahore (170.53 μg/m3), Gujran October and November may be attributed to both cross-border transport
wala (163.63 μg/m3), Okara (139.43 μg/m3), Faisalabad (129.85 μg/ of aerosol produced from biomass burning activities (from India) as well
m3), Pakpattan (126.97 μg/m3), Multan (113.09 μg/m3), Bahawalnagar as locally produced aerosols by anthropogenic activities. As the highest
(110.81 μg/m3), Vehari (110.81 μg/m3), Sargodha (109.81 μg/m3), and values of fine particulates were observed in December and January
Jhelum (107.68 μg/m3). The WHO air quality guidelines (AQG) are not which are not the main months of biomass burning activities, these are
yet defined for PM1 as PM1 is not as widely monitored as PM2.5, there not likely the main source of the high levels of fine particulates pervasive
fore the WHO recommended AQG for PM2.5 (<10 μg/m3 annual mean) across these highly polluted cities. At this time of year, less surface
and Pak-NEQS for PM2.5 (<15 μg/m3 annual mean) are used for com heating and less turbulence due to lower intensity of solar irradiation
parison purposes. Not a single city in Pakistan falls within the PM2.5 lead to stable and shallow boundary layers. Furthermore, with higher
standards defined by Pak-NEQS and WHO, and the values of PM1 and concentrations of light-absorbing aerosols, mainly BC, the atmospheric
PM2.5 respectively for the top 10 cities are 5.6 (8.4) to 9.0 (13.5) times stability increases due to local heating near the top of the boundary
and 7.2 (10.8) to 11.4 (17.1) times greater than the Pak-NEQS (WHO layer, induced by BC, which further lowers the boundary layer height
AQG). For PM1 and PM2.5, 9 out of 10, and 10 out of 10 cities respec (BLH) (Ding et al., 2016). Stable atmospheric conditions that imply low
tively, are in Punjab. The extremely high pollution level may be due to BLH together with low wind speed, both limiting aerosol transport, lead
emissions from local anthropogenic activities, confirming the results of a to the accumulation of aerosols and enhancement of particle concen
previous modeling study that suggested local anthropogenic activities as trations near the surface. As a result, anthropogenic aerosols such as
the major cause of high particulate concentrations in Pakistan (Shi et al., those produced from fossil fuel combustion and other urban and in
2020). All major cities selected in this study (80 cities) are exposed to dustrial activities may linger for long periods (Mhawish et al., 2020). In
PM2.5 concentrations during a long period of time (Figs. 1a and 7a), October and November, both local and remote (cross-border) biomass
which exceed the Pak-NEQS (<15 μg/m3) and 68, 73, and 80, out of 80 (crop residue) burning activities coupled with stable atmospheric con
cities exceeded the WHO Interim Target-1 (<35 μg/m3), Target-2 (<25 ditions have been recognized to cause severe haze and smog episodes,
μg/m3), and Target-3 (<15 μg/m3), respectively. These exceedances are especially over Punjab (Mhawish et al., 2020; Tariq et al., 2015; Tariq
set in strong perspective against the much lower recommended WHO et al., 2016). The formation of secondary inorganic aerosol during haze
AQG for PM2.5 of 10 μg/m3. These results suggest that the top polluted episodes is also responsible for higher PM2.5 concentrations as reported
cities are extremely hazardous for human health, as an increase of PM2.5 from recent studies over China (Nichol et al., 2020; Zhang et al., 2018).
by 10 μg/m3 can increase mortality, lung cancer, and cardiopulmonary An increase in PM2.5 concentrations was observed in June and July, and
12
Fig. 6. Ranking of extremely polluted to polluted cities in Pakistan according to annual mean CAMS PM1 concentrations from 2003 to 2020. Where (a) polluted cities
based on yearly mean PM1 averaged over the years 2003–2020, (b) polluted months based on PM1 averaged over the years 2003–2020, and (c) polluted years based
on yearly mean PM1.
PM1 concentrations slightly increased in July. This means that PM2.5 5.2.3. Characterization of extremely polluted cities using PM10
exhibited two peaks: the first in winter and the second in summer, concentrations
whereas a single peak in winter was observed for PM1. The second PM2.5 Fig. 8a shows the ranking of polluted cities according to PM10 con
peak in summer may be attributed to the fine particulates from dust, as centrations. The PM10 fraction with an aerodynamic diameter larger
dust storm activities are very common in Pakistan during summer, as than PM2.5 (PM10-PM2.5), i.e. the mass concentration of coarse particles,
well as local anthropogenic activities. The lower peak of PM2.5 in the mainly originates from natural sources such as desert dust and resus
summer, compared to winter, may be due to the unstable atmospheric pended soil particles. The top 10 most polluted cities according to the
conditions due to the higher surface heating by solar irradiation, leading PM10 concentrations are Lahore (238.9 μg/m3), Gujranwala (229.1 μg/
to the generation of strong turbulence with rising air and thus strong m3), Okara (194.5 μg/m3), Faisalabad (180.6 μg/m3), Pakpattan (177.9
mixing conditions which promote the vertical dispersion of pollutants. μg/m3), Bahawalnagar (160.6 μg/m3), Vehari (160.6 μg/m3), Multan
The annual mean concentrations of PM1 (Fig. 6c) and PM2.5 (Fig. 7c) (157.5 μg/m3), Sargodha (152.3 μg/m3), and Jhelum (149.7 μg/m3).
show strong inter-annual variations with distinct PMx levels and very PM10 concentrations are 1.2 to 11.9 times higher than the WHO AQG for
poor air quality conditions throughout the last two decades. The annual PM10 (20 μg/m3 annual mean) for all the cities shown in Fig. 8a, sug
mean mass concentrations in extremely polluted cities range from 63 gesting that very poor air quality conditions, hazardous for human life,
μg/m3 to 150.19 μg/m3 for PM1 and from 85 μg/m3 to 187.35 μg/m3 for prevail in all Pakistani cities. Overall, the PM10 temporal trend pattern is
PM2.5, which are 4.2 (6.3)–10 (15) and 5.7 (8.5)–12.5 (18.7) times very similar to that for PM2.5, i.e., December is the month with the
greater than the Pak-NEQS (WHO AQG), respectively. highest PM10 concentrations, followed by January. In summer, July is
the most polluted month followed by June (Fig. 8b). Similar to the PM2.5
variations, PM10 also exhibited peaks in both winter and summer. The
13
Fig. 7. As Fig. 6, but for PM2.5.
higher concentrations during the winter months (i.e. December and concentrations exceeding the WHO recommended AQG for PM10 (<20
January) may be due to increased anthropogenic emission activities μg/m3). Overall, these results suggested that Pakistani cities are a severe
along with stable atmospheric conditions (stagnant conditions, and threat to human life due to extremely poor air quality conditions.
shallower boundary layer). Despite the abundance of coarse particulate
matter in spring and summer seasons which are transported from the 5.2.4. PM1/PM2.5 and PM2.5/PM10 ratios
arid and semiarid regions, the strong convection combined with a The PMx ratios are very useful for understanding the contributions
deeper boundary layer enhances the dispersion of the near-surface among particulate size, as revealed by a study in China where PM1
pollutant that decreases the PM10 concentrations along with the wet contributed nearly 80% of PM2.5 (Wang et al., 2015), which would have
deposition during the rainy summer season. The pre-harvest, harvesting, consequences for human health. Over Pakistan, the PM1/PM2.5 (Fig. 9a)
and post-harvest burning activities along with meteorological condi and PM2.5/PM10 (Fig. 9b) ratios are lower than those observed over
tions such as low wind speed and low boundary layer height may China (Wang et al., 2015), indicating lower contributions of PM1 to
contribute to higher surface PM10 levels especially during October and PM2.5 and PM2.5 to PM10. However, the pattern of ratios is similar to that
November as these activities produce both fine (PM1 and PM2.5) and observed for China, i.e., the PM1/PM2.5 ratios are higher than PM2.5/
coarse (PM10) particles as reported by (Jain et al., 2020; Singh et al., PM10 ratios. Relatively higher PM1/PM2.5 ratios (>75%) are observed
2017) over South Asia and by Le Blond et al. (2017) over South Amer from October to March (Fig. 9a), indicating a larger fraction of PM1 in
ican countries. PM2.5 due to more anthropogenic activities. The directly emitted PM1
Similar to the annual mean PM2.5 variations (Fig. 7c), the annual from the automobile and combustion of fossil fuel, and indirectly by
mean PM10 concentrations also show distinct interannual variations for formation from precursor gases, are most likely higher from October to
all cities (Fig. 8c), and severe air pollution levels were observed March, leading to the enhanced PM1/PM2.5 ratio. This also suggests that
throughout the last two decades. According to these findings, Pakistani the PM2.5 concentrations from October to March are driven by emissions
people are not only exposed to long-term PM2.5 but also to PM10 from combustion and secondary aerosols formation (Jain et al., 2020).
14
Fig. 8. As Fig. 6, but for PM10.
However, low PM1/PM2.5 ratios are observed from April to September in 0.90 and 0.70, respectively. The strong linear relationship between PM10
most of the cities, and low ratios during all months are observed in the and PM2.5 (higher r values) suggests common sources of fine and coarse
cities located in Balochistan, indicating a lower contribution of PM1 to particulates compared to PM1 vs PM2.5 relationship. While the higher
PM2.5, which is mainly dominated by the larger particles especially slope values suggest larger contributions of PM1 to PM2.5 than PM2.5 to
during summer (June, July, and August) which not contributed to PM1. PM10. Overall, both the contribution of PM1 to PM2.5 and that of PM2.5 to
Fig. 9b shows large contributions of PM2.5 to PM10 throughout the PM10 are smaller over Pakistan than over China (Wang et al., 2015) as
year with maximum contributions during summer as indicated by the indicated by the PMx ratios (Fig. 9) and slope values (Fig. 10). This might
large PM2.5/PM10 ratios. This suggests that the air quality in these cities be due to a higher contribution of anthropogenic emissions to the PM
is mainly (and significantly) influenced by fine particulates, largely from concentrations in China than in Pakistan; however, other processes may
anthropogenic sources. The large PM2.5/PM10 ratios in Gwadar and also contribute, and unraveling the different contributions requires
Ormara (Fig. 9b), coastal cities in Balochistan, throughout the year more detailed research. Fig. 10a and b show some scattered points,
suggest that also in these coastal cities the PM is dominated by PM2.5 within a red circle or ellipse, which represent the data from May to
particles, which indicates that the PM10 is driven by PM2.5 which is September and these scattered points suggest lower contributions of
highly influenced by anthropogenic sources. Gwadar has the deepest PM1 in PM2.5 and PM2.5 in PM10, as also indicated by low PMx ratios
seaport in the world and the ship-based emissions may be one of the (Fig. 9).
sources of fine anthropogenic particles throughout the year. However,
lower PM2.5/PM10 ratios are observed for other cities located in Balo 5.2.5. Monthly mean temporal trend of PM1, PM2.5, and PM10
chistan, indicating the greater influence of coarse particulates (mainly The month-to-month variations of the multi-year (2003− 2020)
desert dust). monthly mean PM1, PM2.5, and PM10 concentrations for the top 10
Scatter plots of PM1 vs. PM2.5 (Fig. 10a) and PM2.5 vs. PM10 polluted cities are shown in Fig. 11. These cities vary according to
(Fig. 10b) show that the PMx fractions over Pakistan are well-correlated, population growth, the number of automobiles, urbanization, industri
with Pearson's correlation coefficients (r) of 0.95 and 0.99, and slopes of alization, city size, land cover types, and climatic conditions, and PM
15
Fig. 9. (a) Monthly mean ratios of PM1/PM2.5 and (b) PM2.5/PM10.
Fig. 10. Scatter plots between (a) PM1 vs. PM2.5 and (b) PM2.5 vs. PM10. The red solid line represents the regression line and the black dashed line represents the
identity line. The data points in the red circle and ellipse are explained in the text. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
concentrations are expected to behave differently due to these factors. suggesting more local contributions for the summer months of May to
This study follows the hypothesis of our previous study conducted over August. This local contribution during summer may be attributed to the
Hong Kong (Bilal et al., 2019c) i.e., if the PM concentrations have frequent dust/sand storms. Similarly, from October to January, the PM1,
different magnitudes but follow the same temporal pattern at different PM2.5, and PM10 concentrations in Lahore and Gujranwala show similar
locations, they are influenced by local as well as regional contributions. patterns as in other cities, but with higher concentrations, probably
Thus for PM1 concentrations, Fig. 11a shows the same pattern for each of because Lahore and Gujranwala are the largest cities, with consequently
the 10 cities, suggesting that both local and regional sources contribute more transport, fossil fuel, and industrial emissions, and some local and
to PM1 concentrations. For both PM2.5 (Fig. 11b) and PM10 (Fig. 11c), cross-border biomass burning activities in autumn (Ali et al., 2013; Tariq
similar patterns are only evident from September to April, and dissimilar et al., 2015; Tariq et al., 2016), which reinforce the effects of meteo
patterns due to variation in magnitudes are evident from May to August, rological impacts, such as shallower boundary layer height and lower
16
Fig. 11. Multiyear (2003–2020) monthly average variations of PM1, PM2.5, and PM10 concentrations in the corresponding top 10 polluted cities (see legend). Cities
are plotted with the rank of high to low polluted.
17
wind speed, which result in the accumulation of particulate matter near 1998; Richter and Burrows, 2002). NO2 has an adverse effect on health
the surface (Miao et al., 2019; Miao and Liu, 2019; Miao et al., 2018; Qu and contributes to low atmospheric visibility, and poor air quality
et al., 2017; Sun et al., 2019; Wang et al., 2018). conditions (Khokhar et al., 2015; ul–Haq et al. 2014). Pakistan's top ten
polluted cities according to NO2, where we use Tropospheric vertical
column densities (TVCDs) as a proxy, are those with the highest levels of
5.3. OMI vertical column densities of NO2 and SO2 urbanization, vehicle emissions, and industrialization, suggesting
anthropogenic activities to be the major cause. They are Lahore (5.69 ×
5.3.1. Characterization of extremely polluted cities using NO2 data 1015 molecules/cm2), Rawalpindi (3.65 × 1015 molecules/cm2), Islam
NO2 is mainly produced from fossil fuel combustion, industrial abad (3.65 × 1015 molecules/cm2), Karachi (3.60 × 1015 molecules/
emission, automobile emission, biomass burning, natural lightning, and cm2), Gujranwala (3.32 × 1015 molecules/cm2), Sialkot (2.81 × 1015
soil microbe emissions (Cheng et al., 2012; Lee et al., 1997; Olivier et al.,
Fig. 12. Ranking of extremely polluted to polluted cities in Pakistan according to OMI NO2 TVCDs (molecules/cm2) from 2004 to 2019. (a) polluted cities based on
mean NO2, (b) pixel counts, (c) polluted months based on mean NO2, and (d) polluted years based on mean NO2.
18
molecules/cm2), Haripur (2.73 × 1015 molecules/cm2), Okara (2.72 × attributed to emissions of automobiles, industries, and fossil fuel com
1015 molecules/cm2), Faisalabad (2.72 × 1015 molecules/cm2), and bustion, under stable atmospheric conditions. The NO2 atmospheric
Gujrat (2.47 × 1015 molecules/cm2) (Fig. 12a). Similar results are re lifetime is higher in winter than in summer due to higher mixing ratio
ported by Tabinda et al. (2019), Ashraf et al. (2013), and Khanum et al. and less sunlight that initiates the breakdown reaction of NO2; therefore
(2017). In terms of data availability from OMI, Fig. 12b indicates the stays longer in the atmosphere in winter than in summer. A different
largest number of PC available for Lasbela (4168), Awaran (4154), and trend observed for cities located in Balochistan, with higher NO2 in
Panjgur (4140), all located in Balochistan. On a monthly mean basis, summer, could be due to natural lightning as reported by Khokhar et al.
NO2 (Fig. 12c) follows the same patterns as observed for PM1 and PM2.5 (2015). Fig. 12d shows that Lahore, Rawalpindi, Islamabad, and Karachi
concentrations; i.e., higher values in winter, especially for the extremely are polluted in all years from 2004 to 2019, subjecting citizens to long-
polluted cities (Lahore, Rawalpindi, Islamabad, and Karachi), which are term exposure associated with respiratory diseases, otitis media, and
Fig. 13. Ranking of high to low polluted cities in Pakistan according to OMI SO2 VCDs (molecules/cm2) from 2004 to 2019. (a) polluted cities based on mean SO2,
(b) pixel counts, (c) polluted months based on mean SO2, and (d) polluted years based on mean SO2.
19
mortality (Latza et al., 2009). burning, and industrial activities including construction, manufacturing
of cement, ceramic, and bricks, and metals smelting. These anthropo
5.3.2. Characterization of extremely polluted cities using SO2 data genic sources are mainly responsible for NO2, SO2, and PMx (Adrees
Power plants, oil and gas refineries, and metal smelters are the major et al., 2016; Shah et al., 2012; Ur Rehman et al., 2019). Among these
sources of anthropogenic SO2 (Dahiya and Myllyvirta, 2019). In anthropogenic sources, brick kilns industries are considered a major
Fig. 13a, extremely polluted to polluted cities are ranked based on OMI- source. Small-scale traditional brick kilns, located in rural and suburban
derived SO2 vertical column density and the top 10 polluted cities are areas, produce large amounts of gaseous pollutants (NO2, SO2, O3, and
Lahore (10.6 × 1015 molecules/cm2), Mirpur (10.5 × 1015 molecules/ CO) and PMx due to the usage of low-quality fuels including coal, oil,
cm2), Gujranwala (10.3 × 1015 molecules/cm2), Rawalpindi (10.3 × wood, rice straw, rice husk, rubber tires, bagasse, and corncobs (Adrees
1015 molecules/cm2), Islamabad (10.3 × 1015 molecules/cm2), Sialkot et al., 2016; Ishaq et al., 2010). Besides this, the combustion of agri
(10.3 × 1015 molecules/cm2), Gujrat (10.3 × 1015 molecules/cm2), cultural biomass and crop residue burning are also contributing to
Faisalabad (10.3 × 1015 molecules/cm2), Bhimber (10.2 × 1015 mole deteriorating rural and urban air quality (Irfan et al., 2015; Irfan et al.,
cules/cm2), and Jhelum (10.2 × 1015 molecules/cm2). According to the 2014). Irfan et al. (2015) reported that Punjab produced more aerosol
global SO2 emission hotspot database (Dahiya and Myllyvirta, 2019), pollutants than Sindh from crop residue burning and among the crop
five oil power plants near Lahore are the main sources of high SO2 residues, wheat straw is the main contributor of NOx, SO2, CO2, and CO.
emissions over Lahore. The lower number (1080–2520) of successful Pakistan's 23.6 million vehicles emitted 58% of the country's total NO2
SO2 retrievals (Fig. 13b) as compared to NO2 retrievals (Fig. 12b) is emission and 34% is emitted by power plants and industries (Amnesty
attributed to the high noise level in the OMI-retrieved SO2 data. Only the International, 2019; Government of Pakistan, F.D, 2019; UNDP, 2019).
relatively strong SO2 signal over point sources (e.g., power plants, metal Another important source of aerosol pollutants, missed by previous
smelters) can be detected. (Fioletov et al., 2011; Li et al., 2017; Li et al., studies, is the burning of solid waste and street garbage which is a
2020). The temporal variation of the monthly mean SO2 VCDs (Fig. 13C) common practice in Pakistan, even in major urban cities such as
have a pattern similar to that of PM2.5 and NO2 TVCD, with high values Islamabad, Lahore, Rawalpindi, Faisalabad, Gujranwala, Okara, etc. To
in the winter and low in the summer. For the top polluted cities, the high support this statement, some illustrations with references are provided
SO2 observed during November, December, and January may be in the supplementary data (Fig. S2). Fig. 14a to 14d show that deserts
attributed to the power plants and brick kilns (Dahiya and Myllyvirta, (see Fig. 1 for locations) are another source of increasing AOD and PMx
2019; Rahman et al., 2000). Brick kilns are considered as major sources levels in Pakistan. Although local anthropogenic activities are the mains
of SO2 resulting in extremely poor air quality. This is clearly observed source of aerosol pollutants and severe air quality problems in Pakistan,
over Punjab (Adrees et al., 2016; Colbeck et al., 2010; Pervaiz et al., transboundary transport of aerosols may also influence Pakistan's air
2021; Ur Rehman et al., 2019). Therefore, every year during late autumn quality. Contributions of transboundary transport are investigated in
and winter, the government of Pakistan bans these kilns to control section 5.7, using PSCF analyses, integrated with HYSPLIT backward
pollution levels. The SO2 accumulates in the BL during the stable at trajectory analysis and ground-based PM2.5 measurements.
mospheric conditions and shallow BLH at this time of year, in response
to the low solar irradiation resulting in little surface heating and tur 5.5. Relationship of PMx with AOD, NO2, and SO2
bulence mixing. Unlike NO2, the contribution of SO2 to poor air quality
in Pakistani cities varies from year to year, as shown in Fig. 13d. The SO2 AOD provides valuable information about the aerosol loading in the
VCD is higher in 2004, 2008, and 2011 than in other years. The inves atmospheric column, while the PMx represents the aerosol concentra
tigation of the year-to-year variability requires a separate study. tions near the ground. This section assesses how well satellite-based
AOD describes PM1, PM2.5, and PM10 by examining the monthly corre
5.4. Spatial distributions of aerosols and trace gases lation between AOD and PMx. We have also examined the monthly
correlation between PMx and SO2 and NO2 to understand the common
The purpose of this section is to link the spatial distributions of sources that originated mainly from a combustion process. The re
aerosols and trace gases with each other as well as with population lationships between AOD and PMx vary spatially and temporally and are
density, nighttime lights, land cover types (cropland and urban areas), influenced by several factors such as meteorological variables including
and presumed vegetation fire activities. Here, the PMx data are inter boundary layer height and relative humidity, and the vertical distribu
polated using cubic convolution (Keys, 1981) from 0.75◦ grid size to tion of aerosol layer (Li et al., 2016; Mhawish et al., 2021). The linear
0.125◦ grid size to better show the smooth spatial distributions over correlation between AOD and PMx shows a higher correlation coeffi
different administrative units. The spatial distributions of the multi-year cient from October to January (see Fig. 15a) when the atmosphere is
averaged concentrations of aerosols (AOD, PMx) and trace gases (VCDs) stably stratified and the boundary layer is shallow. This suggests that the
(Fig. 14) show that Punjab is the most polluted region of Pakistan, fol AOD and PMx variability are well agreed during the stable atmospheric
lowed by Sindh. It is significant that other environmental data including conditions (from Oct to Jan) and AOD can explain >65% in the PMx
population density (Fig. 14g), VIIRS nighttime lights (Fig. 14h), crop variability. On the other hand, during April and May when the atmo
land (Fig. 14i), and vegetation fires (Fig. 14j) show similar spatial pat sphere is unstable and the boundary layer deeper, the correlation be
terns. It is obvious that vegetation fires would have the same spatial tween AOD and PMx was smaller (r < 0.4). In the rainy season (July to
pattern as cropland, but not obvious that population density and August), the correlation coefficient between AOD and PM10 was found
nighttime lights would have the same pattern. As nighttime lights and higher than PM2.5 and PM1 which may be due to the larger contribution
vegetation fires represent human activities, having the same spatial of coarse dust particles to the total aerosol loading than PM2.5 and PM1.
patterns suggests that the majority of human settlements including The high relative humidity in the summer season enhanced the AOD
urban, suburban and, industrial regions, are inter-mixed with cropland. retrieval due to the hygroscopic growth of aerosol particles. On the other
Interestingly, these coincident spatial distributions (population, night hand, the wash-out of PMx due to precipitation, deeper boundary layer,
time lights, land cover, and fires) correspond to the higher ranges of and strong convection during rainy months leads to a reduction in the
pollutants i.e., AOD > 0.4, PM1 > 20 μg/m3, PM2.5 > 40 μg/m3, PM10 > ground-level PMx concentrations, while the AOD retrieval remains high
60 μg/m3, NO2 > 1.0 × 1015 molecules/cm2, and SO2 > 6.5 × 1015 under cloud-free conditions during the inactive rain phase (Mhawish
molecules/cm2. These results suggested that the primary (directly et al., 2021). The results suggested that using satellite-based AOD to
emitted) and the secondary (gas-to-particles formation) aerosol emis infer the ground-level PMx variability is limited to specific meteoro
sions and trace gases are mainly from local anthropogenic sources such logical conditions such as stable atmospheric conditions and dry sea
as power plants, oil and gas refineries, vehicular emissions, crop residue sons. On the other hand, the weak linear relationship between AOD and
20
Fig. 14. Spatial distributions of yearly mean (a) AOD averaged over the years 2003–2017 (b) PM1 [2003− 2020] (c) PM2.5 [2003–2020], (d) PM10 [2003–2020], (e)
NO2 [2005–2019], (f) SO2 [2005–2019], (g) Population density [2019], (h) VIIRS Nighttime Lights [2012–2019], (i) Land cover types [2019], and (j) Presumed
vegetation fire data [2003–2020].
21
Fig. 15. Relationship from PMx with (a) AOD, (b) NO2, and (c) SO2 from 2004 to 2017.
ground-level PMx concentrations found during unstable conditions in positive trend in PMx was found over most cities, particularly in Punjab,
spring and summer and more influenced by meteorological variables Khyber Pakhtunkhwa, and the Islamabad Capital Territory. The PMx
and atmospheric mixing height. trends found over cities in Punjab range from +0.35 to +1.10 μg/m3
Tropospheric NO2 and SO2 are precursors for the formation of sec yr− 1, +0.42 to +1.52 μg/m3 yr− 1 and + 0.57 to +2.20 μg/m3 yr− 1 for
ondary aerosols which are produced by anthropogenic activities such as PM1, PM2.5 and PM10, respectively. Correspondingly, the AOD trend in
fossil fuel burning and power plants. The strong correlation coefficient Punjab cities was positive, with the strongest increase over Lahore
between PMx vs. SO2 and NO2 in the spring months suggests that (0.008 yr− 1). Over cities in Khyber Pakhtunkhwa and Azad Kashmir, the
photochemical reactions can contribute to the formation of PMx. The AOD trend was also positive, but smaller than in Punjab. The positive
strong correlation in winter suggests that both trace gases NO2 and SO2 trends in PMx and AOD, particularly over cities in Punjab, may be due to
originated from the same emission sources of PMx, mainly domestic increasing aerosol emissions and/or secondary aerosol formation.
heating, industrial activities, and vehicular emissions. While the lower Anthropogenic activities and biomass burning are considered major
correlation in the summer monsoon may be attributed to the higher sources of ultrafine and fine particles (PM1 and PM2.5) over the region
contribution of natural sources of PMx and the deeper boundary layer (Alam et al., 2015; Stone et al., 2010). Anthropogenic activities also
that enhance the dispersion of air pollutants. result in the production of NO2 and SO2 and ~ 91%, and ~ 88% of the
cities the trends in the NO2 and SO2, respectively, are positive. This in
crease in trace gas concentrations would be a further source of increased
5.6. Trends of aerosol and trace gas concentrations
particulate pollution, as trace gases facilitate secondary aerosol forma
tion via gas-to-particle conversion reactions (Seinfeld and Pandis, 1998).
This section presents the annual trends in the six parameters used to
In terms of monthly trends, the common feature is that the statisti
assess the air quality in each city of Pakistan. The annual trends were
cally significant positive trends of PMx were largest during the cold
calculated after removing the seasonality from the monthly mean time
months (November to February), particularly over major Punjab cities
series data which also accounted for temporal autocorrelation. Fig. 16
(Lahore, Faisalabad, and Gujranwala) and Islamabad (Fig. S3). In
shows the magnitude of the trends as Theil-Sen's slope over each indi
contrast, during the summer months, the trends over many cities are
vidual city, for the periods indicated at the top of each Fig. A significant
22
Fig. 16. Annual mean trend in aerosols and trace gas concentrations for (a) AOD, (b) PM1, (c) PM2.5, (d) PM10, (e) NO2, and (f) SO2. The trends were calculated over
different periods of time, which are indicated on top of each figure on the right-hand side, together with the type of species.
negative. The overall positive annual trends indicate that the increase of 5.7. Potential source contribution function (PSCF) analysis
the PMx concentrations in the winter is stronger than the decrease in the
summer. The reasons for these opposing trends are beyond the scope of PSCF analysis was used to identify the potential source areas for
the current study and require further, more detailed investigation. PM2.5 at four receptor cities: Peshawar, Islamabad, Lahore, and Karachi,
for the period from March 2020 to February 2021. 72 h HYSPLIT
backward trajectories were computed for each receptor site, arriving
every 6 h at the height of 500 m above ground level (AGL). The results
23
Fig. 17. Potential source contribution function plots for PM2.5 at seasonal scales from March 2020 to February 2021 for four receptor cities namely, Peshawar,
Islamabad, Lahore, and Karachi (see legend for identification).
were grouped by season as shown in Fig. 17. The results show strong source regions have a rather low PSCF, and are distributed over specific
differences between cities and seasons. Starting with Peshawar, in spring directions to the W (West) into Afghanistan and toward the SE in India,
there are some local sources regions around the city, within a few with few local sources. In the summer, the source regions are similar to
hundreds of km, but also strong contributions from the WNW (West- those in Peshawar, but with low PSCF except for the source regions in
NorthWest) in Afghanistan and from the SE in India. In the summer, the Afghanistan which seem to contribute most to the air pollution in
source regions are mostly located in Pakistan, but with a contribution Islamabad in the summer, but still with moderate PSCF. In the autumn
from sources to the SE (SouthEast), in India. In contrast, in the autumn sources to the W and N dominate with stronger contributions from
the contributions from India are very small but those from Afghanistan, Afghanistan than from the local sources. In the winter, the source re
both to the NW (NorthWest) and W (West) are relatively large. Whereas, gions redistributed over a much larger area than in other seasons, with
in the winter source regions in NW and SE directions (Afghanistan and strong contributions from both local sources and Afghanistan, as well as
India, respectively) are stronger than in other seasons. In Islamabad, not some contributions from India. The situation in Lahore is remarkably
far from Peshawar, the situation is quite different. In the spring, the different, with the strongest contributions from sources inside Pakistan
24
(PJ and KP), some contributions from sources to the SE in India, during NO2 and SO2 concentrations, were highest. The emissions of these trace
all seasons, and in the spring a strong contribution from sources in gases are known to be associated with anthropogenic activities including
Afghanistan. The situation in Karachi is again different, both as regards transport, industrial activities, and power generation. Interestingly,
source regions and seasonal behaviour. The strongest contributions higher levels of AOD, PM1, PM2.5, PM10, NO2, SO2, population density,
come from local sources within a few hundreds of km in Pakistan, except nighttime lights, and vegetation fire activities showed the same spatial
in the summer when all source regions are weak contributors (PSCF pattern as cropland: most of the major cities, as well as rural areas in
<0.2) and almost all located over the ocean. In the spring, source regions Pakistan, are surrounded by cropland and transport of pollutants
to the NW, reaching far into Afghanistan, contribute to the PM2.5 in generated from anthropogenic activities mix with aerosol and trace
Karachi. Oceanic sources also contribute some to the PM2.5 in other gases generated from agricultural activities, biomass burning and nat
seasons. ural aerosols such as dust, to produce a rather smooth mixture of the
In summary, the values of PSCF indicate the regional transport of metrics reported in this study. These findings suggest that Pakistan's
aerosol from source regions in Afghanistan and India (see Fig. 1 for lo extreme air pollution problems are strongly influenced by anthropo
cations). Karachi is influenced by fine dust particles from the Cholistan genic activities within Pakistan. This is also confirmed by the PSCF (>
and Thar deserts (see Fig. 1 for locations). Fig. 17 shows that the PM2.5 in 0.6) analysis based on HYSPLIT air parcel back trajectories and ground-
Lahore, the top polluted city of Pakistan, is mainly influenced by source based PM2.5 concentrations. In addition, meteorological factors have a
areas in Pakistan, during all seasons. This suggests that increases in local strong influence on the occurrence of pollution episodes.
anthropogenic activities play an important role in the worsening of Significant positive trends in the concentrations of AOD, PM1, PM2.5,
Lahore's air quality. Overall, the higher values of PSCF >0.6 identify PM10, NO2, and SO2 were observed from November to February,
potential source areas which are located both inside and outside of particularly over Lahore, Islamabad, Gujranwala, and Faisalabad.
Pakistan, which indicates that the air quality in Pakistan is not only The final remark of this study is that all cities in Pakistan have been
influenced by local sources but also influenced by transport from exposed to long-term PMx, NO2, and SO2 concentrations throughout the
regional sources areas. last two decades. The pollution levels in these cities imply extremely
poor air quality conditions, mainly due to local anthropogenic activities,
6. Conclusions which severely threaten human life. This comprehensive study, based on
long-term multi-source information on aerosols and trace gases may be
In this study, long-term (2003–2020) remote sensing, ground-based, considered a baseline study by the Ministry of Climate Change, Pakistan,
and model simulation datasets were combined to provide the most and other policymakers, to mitigate air pollution problems in Pakistan.
comprehensive and extensive evaluation ever, of air quality conditions
over Pakistan. Long-term spatio-temporal distributions of aerosol pol Declaration of Competing Interest
lutants and trace gases, recent long-term trends at the city level, ranking
of cities in terms of air pollution levels into three categories (extremely The authors declare that they have no known competing financial
polluted, highly polluted, polluted cities), and the potential sources of interests or personal relationships that could have appeared to influence
air pollution across Pakistan were reported. the work reported in this paper.
The highest AOD was observed in the summer months (June to
August), mainly attributed to the hygroscopic growth of aerosol parti
cles during the humid summer season. High AOD levels were also Acknowledgments
observed during cold months (October to January), mainly over biomass
burning affected regions such as Punjab. For PMx and trace gases, the The authors would like to acknowledge NASA's Level-1 and Atmo
highest values were observed during cold months from October to sphere Archive and Distribution System (LAADS) Distributed Active
February, when the atmosphere is stably stratified and the boundary Archive Center (DAAC) (https://ladsweb.modaps.eosdis.nasa.gov/) for
layer is shallow, and emissions from anthropogenic activities and MODIS data, the Copernicus Atmosphere Monitoring Service (CAMS) for
biomass burning are higher than in other seasons. air quality data (PM1, PM2.5, and PM10), the Goddard Earth Science DISC
The CAMS PM2.5 data are in better agreement with ground-based (https://daac.gsfc.nasa.gov/) for OMI data, Principal Investigators of
PM2.5 concentrations than MERRA-2 reanalysis PM2.5 data and were Lahore AERONET site, and the NOAA Air Resources Laboratory (ARL)
therefore used to rank the cities in terms of concentrations of particulate for the provision of the HYSPLIT air parcel back trajectories (https://
matter (PMx). The 18 years average of the PM2.5 concentrations for the www.ready.noaa.gov) used in this publication. The authors are grate
80 cities of Pakistan show that a total of 21 cities fall within the category ful to Mr. Abid Omar, founder of Pakistan Air Quality Initiative (PAQI),
of extremely polluted cities (PM2.5 > 92.84) (namely Punjab: 17, Khyber and gratefully acknowledge the U.S. Department of State for providing
Pakhtunkhwa: 3, Azad Kashmir: 1), 40 cities fall within the category of the open access ground-based PM2.5 data. The authors also acknowledge
highly polluted cities (45.69 < PM2.5 < 92.84) (namely 6 in Punjab, 14 the use of data from NASA's Fire Information for Resource Management
in Sindh, 3 in Balochistan, 13 in Khyber Pakhtunkhwa and 4 in Azad System (FIRMS) (https://earthdata.nasa.gov/firms), part of NASA's
Kashmir); 19 cities fall within the category of polluted cities (PM2.5 < Earth Observing System Data and Information System (EOSDIS). The
45.69) (16 in Balochistan and 3 in Khyber Pakhtunkhwa). No single city authors are thankful to Mr. Pravash Tiwari for helping in PSCF analysis
in Pakistan falls within the PM2.5 standards defined by Pak-NEQS and and Dr. Devin White (Oak Ridge National Laboratory) for MODIS Con
WHO, and the values of PM1 and PM2.5 for the top 10 cities are 5.6 (8.4) version Tool Kit (MCTK).
to 9.0 (13.5) times and 7.2 (10.8) to 11.4 (17.1) times larger than the
Pak-NEQS (WHO AQG). The map of annual average PM2.5 shows that Funding
people in the whole country are exposed to high PM2.5 concentrations
for many years, with the annual mean concentrations for all cities This work was supported by the National Key Research and Devel
exceeding the Pak-NEQS (<15 μg/m3), and 68, 73, and 80 cities opment Program of China (2016YFC1400901), the Special Project of
exceeding the WHO Interim Target-1 (<35 μg/m3), Target-2 (<25 μg/ Jiangsu Distinguished Professor (R2018T22), the National Natural Sci
m3), and Target-3 (<15 μg/m3), respectively. In terms of pollution ence Foundation of China (41976165), Jiangsu Technology Project of
sources, the study indicates that biomass (crop residue) burning activ Nature Resources (KJXM2019042), and the Startup Foundation for
ities may not be the main source of severe air quality conditions in Introduction Talent of NUIST (2017r107). Additional support came from
Pakistan: the highest PMx concentrations were observed in December the New Mexico State University College of Agriculture Consumer and
and January when also the NO2 TVCD and SO2 VCD, used as proxies for Environmental Sciences' Agricultural Experiment Station.
25
Appendix A. Supplementary data Colarco, P., da Silva, A., Chin, M., Diehl, T., 2010. Online simulations of global aerosol
distributions in the NASA GEOS-4 model and comparisons to satellite and ground-
based aerosol optical depth. J. Geophys. Res. 115.
Supplementary data to this article can be found online at https://doi. Colbeck, I., Nasir, Z.A., Ali, Z., 2010. The state of ambient air quality in Pakistan–a
org/10.1016/j.rse.2021.112617. review. Environ. Sci. Pollut. Res. Int. 17, 49–63.
Cuevas, E., Camino, C., Benedetti, A., Basart, S., Terradellas, E., Baldasano, J.M.,
Morcrette, J.J., Marticorena, B., Goloub, P., Mortier, A., Berjón, A., Hernández, Y.,
References Gil-Ojeda, M., Schulz, M., 2015. The MACC-II 2007–2008 reanalysis: atmospheric
dust evaluation and characterization over northern Africa and the Middle East.
Adrees, M., Ibrahim, M., Shah, A.M., Abbas, F., Saleem, F., Rizwan, M., Hina, S., Atmos. Chem. Phys. 15, 3991–4024.
Jabeen, F., Ali, S., 2016. Gaseous pollutants from brick kiln industry decreased the Dahiya, S., Myllyvirta, L., 2019. Global SO2 emission hotspot database: ranking the
growth, photosynthesis, and yield of wheat (Triticum aestivum L.). Environ. Monit. world’s worst sources of SO2 pollution. K. Ford, N. Sivalingam, S. Ayech, & A.
Assess. 188, 267. Jacobsen (Eds.) https://www.greenpeace.org/static/planet4-africa-stateless/2019/
Alam, K., Rahman, N., Khan, H.U., Haq, B.S., Rahman, S., 2015. Particulate matter and 08/5f139f4c-final-global-hotspot-and-emission-sources-for-so2_19th_august-2019.
its source apportionment in Peshawar, northern Pakistan. Aerosol Air Qual. Res. 15, pdf, access date: 23 March 2021.
634–647. de Leeuw, G., Sogacheva, L., Rodriguez, E., Kourtidis, K., Georgoulias, A.K.,
Ali, M.A., Assiri, M., 2019. Analysis of AOD from MODIS-merged DT–DB products over Alexandri, G., Amiridis, V., Proestakis, E., Marinou, E., Xue, Y., van der A, R., 2018.
the Arabian Peninsula. Earth Syst. Environ. 3, 625–636. Two decades of satellite observations of AOD over mainland China using ATSR-2,
Ali, M., Tariq, S., Mahmood, K., Daud, A., Batool, A., Zia ul, H., 2013. A study of aerosol AATSR and MODIS/Terra: data set evaluation and large-scale patterns. Atmos.
properties over Lahore (Pakistan) by using AERONET data. Asia-Pac. J. Atmos. Sci. Chem. Phys. 18, 1573–1592.
50, 153–162. Dickerson, R.R., Kondragunta, S., Stenchikov, G., Civerolo, K.L., Doddridge, B.G.,
Amnesty International, 2019. Pakistan: hazardous air puts lives at risk https://www. Holben, B.N., 1997. The impact of aerosols on solar ultraviolet radiation and
amnesty.org/en/latest/news/2019/10/pakistan-hazardous-air/, access date: 1 June. photochemical smog. Science 278, 827–830.
Ashraf, N., Mushtaq, M., Sultana, B., Iqbal, M., Ullah, I., Shahid, S.A., 2013. Preliminary Ding, A.J., Huang, X., Nie, W., Sun, J.N., Kerminen, V.M., Petaja, T., Su, H., Cheng, Y.F.,
monitoring of tropospheric air quality of Lahore City in Pakistan. Int. J. Chem. Yang, X.Q., Wang, M.H., Chi, X.G., Wang, J.P., Virkkula, A., Guo, W.D., Yuan, J.,
Biochem. Sci. 3, 19–28. Wang, S.Y., Zhang, R.J., Wu, Y.F., Song, Y., Zhu, T., Zilitinkevich, S., Kulmala, M.,
Begum, A.B., Kim, E., Jeong, C.-H., Lee, D.-W., Hopke, P.K., 2005. Evaluation of the Fu, C.B., 2016. Enhanced haze pollution by black carbon in megacities in China.
potential source contribution function using the 2002 Quebec forest fire episode. Geophys. Res. Lett. 43, 2873–2879.
Atmos. Environ. 39, 3719–3724. Eck, T.F., Holben, B.N., Reid, J.S., Dubovik, O., Smirnov, A., O'Neill, N.T., Slutsker, I.,
Bilal, M., Nichol, J., 2017. Evaluation of the NDVI-based pixel selection criteria of the Kinne, S., 1999. Wavelength dependence of the optical depth of biomass burning,
MODIS C6 dark target and deep blue combined aerosol product. IEEE J. Select. Top. urban, and desert dust aerosols. J. Geophys. Res.-Atmos. 104, 31333–31349.
Appl. Earth Observ. Remote Sens. 10, 3448–3453. Elvidge, C.D., Zhizhin, M., Ghosh, T., Hsu, F.-C., Taneja, J., 2021. Annual time series of
Bilal, M., Nichol, J.E., Bleiweiss, M.P., Dubois, D., 2013. A simplified high resolution global VIIRS nighttime lights derived from monthly averages: 2012 to 2019. Remote
MODIS aerosol retrieval algorithm (SARA) for use over mixed surfaces. Remote Sens. Sens. 13.
Environ. 136, 135–145. Fan, A., Chen, W., Liang, L., Sun, W., Lin, Y., Che, H., Zhao, X., 2017. Evaluation and
Bilal, M., Nichol, J.E., Chan, P.W., 2014. Validation and accuracy assessment of a comparison of long-term MODIS C5.1 and C6 products against AERONET
simplified aerosol retrieval algorithm (SARA) over Beijing under low and high observations over China. Remote Sens. 9, 1269.
aerosol loadings and dust storms. Remote Sens. Environ. 153, 50–60. Filonchyk, M., Yan, H., Zhang, Z., Yang, S., Li, W., Li, Y., 2019. Combined use of satellite
Bilal, M., Nichol, J.E., Nazeer, M., 2016. Validation of aqua-MODIS C051 and C006 and surface observations to study aerosol optical depth in different regions of China.
operational aerosol products using AERONET measurements over Pakistan. IEEE J. Sci. Rep. 9, 6174.
Select. Top. Appl. Earth Observ. Remote Sens. 9, 2074–2080. Fioletov, V.E., McLinden, C.A., Krotkov, N., Moran, M.D., Yang, K., 2011. Estimation of
Bilal, M., Nichol, J., Wang, L., 2017. New customized methods for improvement of the SO2emissions using OMI retrievals. Geophys. Res. Lett. 38 n/a-n/a.
MODIS C6 dark target and deep blue merged aerosol product. Remote Sens. Environ. Fleming, Z.L., Monks, P.S., Manning, A.J., 2012. Review: untangling the influence of air-
197, 115–124. mass history in interpreting observed atmospheric composition. Atmos. Res. 104-
Bilal, M., Nazeer, M., Qiu, Z., Ding, X., Wei, J., 2018a. Global validation of MODIS C6 105, 1–39.
and C6.1 merged aerosol products over diverse vegetated surfaces. Remote Sens. 10. Flemming, J., Huijnen, V., Arteta, J., Bechtold, P., Beljaars, A., Blechschmidt, A.M.,
Bilal, M., Qiu, Z., Campbell, J.R., Spak, S., Shen, X., Nazeer, M., 2018b. A new MODIS C6 Diamantakis, M., Engelen, R.J., Gaudel, A., Inness, A., Jones, L., Josse, B.,
dark target and deep blue merged aerosol product on a 3 km spatial grid. Remote Katragkou, E., Marecal, V., Peuch, V.H., Richter, A., Schultz, M.G., Stein, O.,
Sens. 10, 463. Tsikerdekis, A., 2015. Tropospheric chemistry in the integrated forecasting system of
Bilal, M., Nazeer, M., Nichol, J.E., Bleiweiss, M.P., Qiu, Z., Jäkel, E., Campbell, J.R., ECMWF. Geosci. Model Dev. 8, 975–1003.
Atique, L., Huang, X., Lolli, S., 2019a. A simplified and robust surface reflectance Flemming, J., Benedetti, A., Inness, A., Engelen, R.J., Jones, L., Huijnen, V., Remy, S.,
estimation method (SREM) for use over diverse land surfaces using multi-sensor Parrington, M., Suttie, M., Bozzo, A., Peuch, V.-H., Akritidis, D., Katragkou, E., 2017.
data. Remote Sens. 11. The CAMS interim reanalysis of carbon monoxide, ozone and aerosol for 2003–2015.
Bilal, M., Nazeer, M., Nichol, J.E., Qiu, Z., Wang, L., Bleiweiss, M.P., Shen, X., Atmos. Chem. Phys. 17, 1945–1983.
Campbell, J.R., Lolli, S., 2019b. Evaluation of terra-MODIS C6 and C6.1 aerosol Giles, D.M., Sinyuk, A., Sorokin, M.G., Schafer, J.S., Smirnov, A., Slutsker, I., Eck, T.F.,
products against Beijing, XiangHe, and Xinglong Aeronet Sites in China during 2004- Holben, B.N., Lewis, J.R., Campbell, J.R., Welton, E.J., Korkin, S.V., Lyapustin, A.I.,
2014. Remote Sens. 11, 486. 2019. Advancements in the aerosol Robotic network (AERONET) version 3 database
Bilal, M., Nichol, J.E., Nazeer, M., Shi, Y., Wang, L., Kumar, K.R., Ho, H.C., Mazhar, U., – automated near-real-time quality control algorithm with improved cloud screening
Bleiweiss, M.P., Qiu, Z., Khedher, K.M., Lolli, S., 2019c. Characteristics of fine for Sun photometer aerosol optical depth (AOD) measurements. Atmos. Measur.
particulate matter (PM2.5) over urban, suburban, and rural areas of Hong Kong. Tech. 12, 169–209.
Atmosphere 10. Government of Pakistan, F.D, 2019. Pakistan Economic Survey 2018-19 https://www.
Buchard, V., da Silva, A.M., Randles, C.A., Colarco, P., Ferrare, R., Hair, J., Hostetler, C., finance.gov.pk/survey_1819.html. access date: 1 June 2020.
Tackett, J., Winker, D., 2016. Evaluation of the surface PM2.5 in version 1 of the Granier, C., Bessagnet, B., Bond, T., D’Angiola, A., Denier van der Gon, H., Frost, G.J.,
NASA MERRA aerosol reanalysis over the United States. Atmos. Environ. 125, Heil, A., Kaiser, J.W., Kinne, S., Klimont, Z., Kloster, S., Lamarque, J.-F., Liousse, C.,
100–111. Masui, T., Meleux, F., Mieville, A., Ohara, T., Raut, J.-C., Riahi, K., Schultz, M.G.,
Buchard, V., Randles, C.A., da Silva, A.M., Darmenov, A., Colarco, P.R., Govindaraju, R., Smith, S.J., Thompson, A., van Aardenne, J., van der Werf, G.R., van Vuuren, D.P.,
Ferrare, R., Hair, J., Beyersdorf, A.J., Ziemba, L.D., Yu, H., 2017. The MERRA-2 2011. Evolution of anthropogenic and biomass burning emissions of air pollutants at
aerosol reanalysis, 1980 onward. Part II: evaluation and case studies. J. Clim. 30, global and regional scales during the 1980–2010 period. Clim. Chang. 109, 163–190.
6851–6872. Gupta, P., Khan, M.N., da Silva, A., Patadia, F., 2013. MODIS aerosol optical depth
Carn, S.A., Fioletov, V.E., McLinden, C.A., Li, C., Krotkov, N.A., 2017. A decade of global observations over urban areas in Pakistan: quantity and quality of the data for air
volcanic SO2 emissions measured from space. Sci. Rep. 7, 44095. quality monitoring. Atmos. Pollut. Res. 4, 43–52.
Che, H., Yang, L., Liu, C., Xia, X., Wang, Y., Wang, H., Wang, H., Lu, X., Zhang, X., 2019. Hamed, K.H., Ramachandra Rao, A., 1998. A modified Mann-Kendall trend test for
Long-term validation of MODIS C6 and C6.1 dark target aerosol products over China autocorrelated data. J. Hydrol. 204, 182–196.
using CARSNET and AERONET. Chemosphere 236, 124268. Harper, W.V., 2016. Reduced Major Axis Regression. Wiley StatsRef: Statistics Reference
Chen, G., Li, S., Zhang, Y., Zhang, W., Li, D., Wei, X., He, Y., Bell, M.L., Williams, G., Online, pp. 1–6.
Marks, G.B., Jalaludin, B., Abramson, M.J., Guo, Y., 2017. Effects of ambient PM 1 He, L., Wang, L., Lin, A., Zhang, M., Bilal, M., Wei, J., 2018. Performance of the NPP-
air pollution on daily emergency hospital visits in China: an epidemiological study. VIIRS and aqua-MODIS aerosol optical depth products over the Yangtze River basin.
Lancet Planetary Health 1, e221–e229. Remote Sens. 10, 117.
Cheng, M.M., Jiang, H., Guo, Z., 2012. Evaluation of long-term tropospheric NO2 He, L., Lin, A., Chen, X., Zhou, H., Zhou, Z., He, P., 2019. Assessment of MERRA-2 surface
columns and the effect of different ecosystem in Yangtze River Delta. Procedia PM2.5 over the Yangtze River basin: ground-based verification, spatiotemporal
Environ. Sci. 13, 1045–1056. distribution and meteorological dependence. Remote Sens. 11.
Chin, M., Ginoux, P., Kinne, S., Torres, O., Holben, B.N., Duncan, B.N., Martin, R.V., Health Effects Institute, 2019. State of Global Air 2019: A Special Report on Global
Logan, J.A., Higurashi, A., Nakajima, T., 2002. Tropospheric aerosol optical Exposure to Air Pollution and Its Disease Burden. In. Boston, MA.
thickness from the GOCART model and comparisons with satellite and sun Holben, B.N., Eck, T.F., Slutsker, I., Tanré, D., Buis, J.P., Setzer, A., Vermote, E.,
photometer measurements. J. Atmos. Sci. 59, 461–483. Reagan, J.A., Kaufman, Y.J., Nakajima, T., Lavenu, F., Jankowiak, I., Smirnov, A.,
26
1998. AERONET—A federated instrument network and data archive for aerosol Li, L., Wu, J., Ghosh, J.K., Ritz, B., 1994. Estimating spatiotemporal variability of
characterization. Remote Sens. Environ. 66, 1–16. ambient air pollutant concentrations with A hierarchical model. Atmos. Environ. 71,
Holben, N., Tanr, D., Smirnov, A., Eck, T.F., Slutsker, I., Newcomb, W.W., Schafer, J.S., 54–63.
Chatenet, B., Lavenu, F., Kaufman, J., Castle, J.V., Setzer, A., Markham, B., Clark, D., Li, L., Wang, Y., 2014. What drives the aerosol distribution in Guangdong–the most
Halthore, R., Karneli, A., Neill, N.T.O., Pietras, C., Pinker, T., Voss, K., Zibordi, G., developed province in southern China? Sci. Rep. 4, 5972.
2001. An emerging ground-based aerosol climatology : aerosol optical depth from Li, Z.Q., Zhang, Y., Shao, J., Li, B.S., Hong, J., Liu, D., Li, D.H., Wei, P., Li, W., Li, L.,
AERONET. J. Geophys. Res.-Atmos. 106, 12067–12097. Zhang, F.X., Guo, J., Deng, Q., Wang, B.X., Cui, C.L., Zhang, W.C., Wang, Z.Z., Lv, Y.,
Hsu, N.C., Jeong, M.-J., Bettenhausen, C., Sayer, A.M., Hansell, R., Seftor, C.S., Huang, J., Xu, H., Chen, X.F., Li, L., Qie, L.L., 2016. Remote sensing of atmospheric particulate
Tsay, S.-C., 2013. Enhanced deep blue aerosol retrieval algorithm: the second mass of dry PM2.5 near the ground: method validation using ground-based
generation. J. Geophys. Res.-Atmos. 118, 9296–9315. measurements. Remote Sens. Environ. 173, 59–68.
Inness, A., Ades, M., Agustí-Panareda, A., Barré, J., Benedictow, A., Blechschmidt, A.-M., Li, C., Krotkov, N.A., Carn, S., Zhang, Y., Spurr, R.J.D., Joiner, J., 2017. New-generation
Dominguez, J.J., Engelen, R., Eskes, H., Flemming, J., Huijnen, V., Jones, L., NASA Aura Ozone Monitoring Instrument (OMI) volcanic SO2 dataset: algorithm
Kipling, Z., Massart, S., Parrington, M., Peuch, V.-H., Razinger, M., Remy, S., description, initial results, and continuation with the Suomi-NPP Ozone Mapping
Schulz, M., Suttie, M., 2019. The CAMS reanalysis of atmospheric composition. and Profiler Suite (OMPS). Atmos. Measur. Tech. 10, 445–458.
Atmos. Chem. Phys. 19, 3515–3556. Li, C., Krotkov, N.A., Leonard, P.J.T., Carn, S., Joiner, J., Spurr, R.J.D., Vasilkov, A.,
Irfan, M., Riaz, M., Arif, M.S., Shahzad, S.M., Saleem, F., Rahman, N.-u., van den Berg, L., 2020. Version 2 Ozone Monitoring Instrument SO2 product (OMSO2
& Abbas, F., 2014. Estimation and characterization of gaseous pollutant emissions V2): new anthropogenic SO2 vertical column density dataset. Atmos.
from agricultural crop residue combustion in industrial and household sectors of Measur. Tech. 13, 6175–6191.
Pakistan. Atmos. Environ. 84, 189–197. Liu, L., Breitner, S., Schneider, A., Cyrys, J., Bruske, I., Franck, U., Schlink, U., Marian
Irfan, M., Riaz, M., Arif, M.S., Shahzad, S.M., Hussain, S., Akhtar, M.J., van den Berg, L., Leitte, A., Herbarth, O., Wiedensohler, A., Wehner, B., Pan, X., Wichmann, H.E.,
Abbas, F., 2015. Spatial distribution of pollutant emissions from crop residue Peters, A., 2013. Size-fractioned particulate air pollution and cardiovascular
burning in the Punjab and Sindh provinces of Pakistan: uncertainties and challenges. emergency room visits in Beijing, China. Environ. Res. 121, 52–63.
Environ. Sci. Pollut. Res. Int. 22, 16475–16491. Livingston, J.M., Redemann, J., Shinozuka, Y., Johnson, R., Russell, P.B., Zhang, Q.,
Ishaq, M., Khan, M.A., Jan, F.A., Ahmad, I., 2010. Heavy metals in brick kiln located area Mattoo, S., Remer, L., Levy, R., Munchak, L., Ramachandran, S., 2014. Comparison
using atomic absorption spectrophotometer: a case study from the city of Peshawar, of MODIS 3 km and 10 km resolution aerosol optical depth retrievals over land with
Pakistan. Environ. Monit. Assess. 166, 409–420. airborne sunphotometer measurements during ARCTAS summer 2008. Atmos.
Islam, M.N., Ali, M.A., Islam, M.M., 2019. Spatiotemporal investigations of aerosol Chem. Phys. 14, 2015–2038.
optical properties over Bangladesh for the period 2002–2016. Earth Syst. Environ. 3, Mann, H.B., 1945. Nonparametric tests against trend. Econometrica 13.
563–573. Mannucci, P.M., Franchini, M., 2017. Health effects of ambient air pollution in
Jacob, D.J., Winner, D.A., 2009. Effect of climate change on air quality. Atmos. Environ. developing countries. Int. J. Environ. Res. Public Health 14.
43, 51–63. Mehta, M., Singh, R., Singh, A., Singh, N., Anshumali, 2016. Recent global aerosol optical
Jain, S., Sharma, S.K., Vijayan, N., Mandal, T.K., 2020. Seasonal characteristics of depth variations and trends - a comparative study using MODIS and MISR level 3
aerosols (PM2.5 and PM10) and their source apportionment using PMF: A four year datasets. Remote Sens. Environ. 181, 137–150.
study over Delhi, India. Environ. Pollut. 262, 114337. Mei, L., Zhao, C., de Leeuw, G., Che, H., Che, Y., Rozanov, V., Vountas, M., Burrows, J.P.,
Janssens-Maenhout, G., Pagliari, V., Guizzardi, D., Muntean, M., 2013. Global Emission 2019. Understanding MODIS dark-target collection 5 and 6 aerosol data over China:
Inventories in the Emission Database for Global Atmospheric Research (EDGAR) – effect of surface type, aerosol loading and aerosol absorption. Atmos. Res. 228,
Manual (I), Gridding: EDGAR emissions distribution on global gridmaps. 161–175.
Publications Office of the European Union, Luxembourg. Meng, X., Ma, Y., Chen, R., Zhou, Z., Chen, B., Kan, H., 2013. Size-fractionated particle
Jethva, H., Torres, O., Field, R.D., Lyapustin, A., Gautam, R., Kayetha, V., 2019. number concentrations and daily mortality in a Chinese city. Environ. Health
Connecting crop productivity, residue fires, and air quality over northern India. Sci. Perspect. 121, 1174–1178.
Rep. 9, 16594. Mhawish, A., Banerjee, T., Broday, D.M., Misra, A., Tripathi, S.N., 2017. Evaluation of
Kendall, M., Gibbons, J.D., 1990. Rank Correlation Methods, 5th Edition ed. Edward MODIS collection 6 aerosol retrieval algorithms over indo-Gangetic plain:
Arnold, London. implications of aerosols types and mass loading. Remote Sens. Environ. 201,
Keys, R., 1981. Cubic convolution interpolation for digital image processing. IEEE Trans. 297–313.
Acoust. Speech Signal Process. 29, 1153–1160. Mhawish, A., Banerjee, T., Sorek-Hamer, M., Lyapustin, A., Broday, D.M., Chatfield, R.,
Khanum, F., Chaudhry, M.N., Kumar, P., 2017. Characterization of five-year observation 2019. Comparison and evaluation of MODIS multi-angle implementation of
data of fine particulate matter in the metropolitan area of Lahore. Air Qual. Atmos. atmospheric correction (MAIAC) aerosol product over South Asia. Remote Sens.
Health 10, 725–736. Environ. 224, 12–28.
Khokhar, M.F., Yasmin, N., Fatima, N., Beirle, S., Wagner, T., 2015. Detection of trends Mhawish, A., Banerjee, T., Sorek-Hamer, M., Bilal, M., Lyapustin, A.I., Chatfield, R.,
and seasonal variation in tropospheric nitrogen dioxide over Pakistan. Aerosol Air Broday, D.M., 2020. Estimation of high-resolution PM2.5 over the indo-Gangetic
Qual. Res. 15, 2508–2524. plain by fusion of satellite data, meteorology, and land use variables. Environ. Sci.
Khokhar, M.F., Mehdi, H., Abbas, Z., Javed, Z., 2016. Temporal assessment of NO2 Technol. 54, 7891–7900.
pollution levels in urban centers of Pakistan by employing ground-based and satellite Mhawish, A., Sorek-Hamer, M., Chatfield, R., Banerjee, T., Bilal, M., Kumar, M.,
observations. Aerosol Air Qual. Res. 16, 1854–1867. Sarangi, C., Franklin, M., Chau, K., Garay, M., Kalashnikova, O., 2021. Aerosol
Krotkov, N.A., McLinden, C.A., Li, C., Lamsal, L.N., Celarier, E.A., Marchenko, S.V., characteristics from earth observation systems: a comprehensive investigation over
Swartz, W.H., Bucsela, E.J., Joiner, J., Duncan, B.N., Boersma, K.F., Veefkind, J.P., South Asia (2000-2019). Remote Sens. Environ. 259, 112410 https://doi.org/
Levelt, P.F., Fioletov, V.E., Dickerson, R.R., He, H., Lu, Z., Streets, D.G., 2016. Aura 10.1016/j.rse.2021.112410.
OMI observations of regional SO2 and NO2 pollution changes from 2005 to 2015. Miao, Y., Liu, S., 2019. Linkages between aerosol pollution and planetary boundary layer
Atmos. Chem. Phys. 16, 4605–4629. structure in China. Sci. Total Environ. 650, 288–296.
Krotkov, N.A., Lamsal, L.N., Celarier, E.A., Swartz, W.H., Marchenko, S.V., Bucsela, E.J., Miao, Y., Liu, S., Guo, J., Huang, S., Yan, Y., Lou, M., 2018. Unraveling the relationships
Chan, K.L., Wenig, M., Zara, M., 2017. The version 3 OMI NO2 between boundary layer height and PM2.5 pollution in China based on four-year
standard product. Atmos. Measur. Tech. 10, 3133–3149. radiosonde measurements. Environ. Pollut. 243, 1186–1195.
Kulmala, M., Dada, L., Daellenbach, K.R., Yan, C., Stolzenburg, D., Kontkanen, J., Miao, Y., Li, J., Miao, S., Che, H., Wang, Y., Zhang, X., Zhu, R., Liu, S., 2019. Interaction
Ezhova, E., Hakala, S., Tuovinen, S., Kokkonen, T.V., Kurppa, M., Cai, R., Zhou, Y., between planetary boundary layer and PM2.5 pollution in megacities in China: a
Yin, R., Baalbaki, R., Chan, T., Chu, B., Deng, C., Fu, Y., Ge, M., He, H., Heikkinen, L., review. Curr. Pollut. Rep. 5, 261–271.
Junninen, H., Liu, Y., Lu, Y., Nie, W., Rusanen, A., Vakkari, V., Wang, Y., Yang, G., More, S., Kumar, P.P., Gupta, P., Devara, P., Aher, G., 2013. Comparison of aerosol
Yao, L., Zheng, J., Kujansuu, J., Kangasluoma, J., Petaja, T., Paasonen, P., Jarvi, L., products retrieved from AERONET, MICROTOPS and MODIS over a tropical Urban
Worsnop, D., Ding, A., Liu, Y., Wang, L., Jiang, J., Bianchi, F., Kerminen, V.M., 2020. City, Pune, India. Aerosol Air Qual. Res. 13, 107–121.
Is Reducing New Particle Formation a Plausible Solution to Mitigate Particulate Air Navinya, C.D., Vinoj, V., Pandey, S.K., 2020. Evaluation of PM2.5 surface concentrations
Pollution in Beijing and Other Chinese Megacities? Faraday Discuss. simulated by NASA’s MERRA version 2 aerosol reanalysis over India and its relation
Latza, U., Gerdes, S., Baur, X., 2009. Effects of nitrogen dioxide on human health: to the air quality index. Aerosol Air Qual. Res. 20, 1329–1339.
systematic review of experimental and epidemiological studies conducted between Nichol, J., Bilal, M., 2016. Validation of MODIS 3 km resolution aerosol optical depth
2002 and 2006. Int. J. Hyg. Environ. Health 212, 271–287. retrievals over Asia. Remote Sens. 8, 328.
Le Blond, J.S., Woskie, S., Horwell, C.J., Williamson, B.J., 2017. Particulate matter Nichol, J.E., Bilal, M., Ali, M.A., Qiu, Z., 2020. Air pollution scenario over China during
produced during commercial sugarcane harvesting and processing: a respiratory COVID-19. Remote Sens. 12.
health hazard? Atmos. Environ. 149, 34–46. Olivier, J.G.J., Bouwman, A.F., Van der Hoek, K.W., Berdowski, J.J.M., 1998. Global air
Lee, D.S., Köhler, I., Grobler, E., Rohrer, F., Sausen, R., Gallardo-Klenner, L., Olivier, J.G. emission inventories for anthropogenic sources of NOx, NH3 and N2O in 1990.
J., Dentener, F.J., Bouwman, A.F., 1997. Estimations of global no, emissions and Environ. Pollut. 102, 135–148.
their uncertainties. Atmos. Environ. 31, 1735–1749. PAQI, 2018. LahoreSmog, Just how Bad Is it? Pakistan Air Quality Initiative, In.
Levy, R.C., Remer, L.a., Kleidman, R.G., Mattoo, S., Ichoku, C., Kahn, R., Eck, T.F., 2010. Pervaiz, S., Akram, M.A.N., Khan, F.Z., Javid, K., Zahid, Y., 2021. Brick sector and air
Global evaluation of the collection 5 MODIS dark-target aerosol products over land. quality: an integrated assessment towards 2020 challenge of environment
Atmos. Chem. Phys. 10, 10399–10420. development. Environ. Natl Reso. J. 19, 153–164.
Levy, R.C., Mattoo, S., Munchak, L.A., Remer, L.A., Sayer, A.M., Patadia, F., Hsu, N.C., Pope, C.A., Burnett, R.T., Thun, M.J., Calle, E.E., Krewski, D., Ito, K., Thurston, G.D.,
2013. The collection 6 MODIS aerosol products over land and ocean. Atmos. Measur. 2002. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine
Tech. 6, 2989–3034. particulate air pollution. JAMA : J. Am. Med. Assoc. 287, 1132–1141.
27
Provencal, S., Buchard, V., da Silva, A.M., Leduc, R., Barrette, N., Elhacham, E., Wang, S. Stone, E., Schauer, J., Quraishi, T.A., Mahmood, A., 2010. Chemical characterization and
H., 2017. Evaluation of PM2.5 surface concentration simulated by version 1 of the source apportionment of fine and coarse particulate matter in Lahore, Pakistan.
NASA’s MERRA aerosol reanalysis over Israel and Taiwan. Aerosol Air Qual. Res. 17, Atmos. Environ. 44, 1062–1070.
253–261. Sun, T., Che, H., Qi, B., Wang, Y., Dong, Y., Xia, X., Wang, H., Gui, K., Zheng, Y.,
Purohit, P., Munir, T., Rafaj, P., 2013. Scenario analysis of strategies to control air Zhao, H., Ma, Q., Du, R., Zhang, X., 2019. Characterization of vertical distribution
pollution in Pakistan. J. Integr. Environ. Sci. 10, 77–91. and radiative forcing of ambient aerosol over the Yangtze River Delta during 2013-
Qu, Y., Han, Y., Wu, Y., Gao, P., Wang, T., 2017. Study of PBLH and its correlation with 2015. Sci. Total Environ. 650, 1846–1857.
particulate matter from one-year observation over Nanjing, Southeast China. Remote Tabinda, A.B., Ali, H., Yasar, A., Rasheed, R., Mahmood, A., Iqbal, A., 2019. Comparative
Sens. 9. assessment of ambient air quality of major cities of Pakistan. Mapan 35, 25–32.
Rahman, U., Awan, M.A., Hassan, S.T., Khattak, M.M., 2000. Mosses as indicators of Tariq, S., Ali, M., 2015. Spatio–temporal distribution of absorbing aerosols over Pakistan
atmospheric pollution of trace metals (Cd, Cu, Pb, Mn and Zn) in the vicinity of coal- retrieved from OMI onboard Aura satellite. Atmos. Pollut. Res. 6, 254–266.
fired brick kilns in north-eastern suburbs of Islamabad, Pakistan. J. Radioanal. Nucl. Tariq, S., ul-Haq, Z., Ali, M., 2015. Analysis of optical and physical properties of aerosols
Chem. 246, 331–336. during crop residue burning event of October 2010 over Lahore, Pakistan. Atmos.
Randles, C.A., Da Silva, A.M., Buchard, V., Colarco, P.R., Darmenov, A., Govindaraju, R., Pollut. Res. 6, 969–978.
Smirnov, A., Holben, B., Ferrare, R., Hair, J., Shinozuka, Y., Flynn, C.J., 2017. The Tariq, S., Zia, U.-H., Ali, M., 2016. Satellite and ground-based remote sensing of aerosols
MERRA-2 aerosol reanalysis, 1980 - onward, part I: system description and data during intense haze event of October 2013 over Lahore, Pakistan. Asia-Pac. J. Atmos.
assimilation evaluation. J. Clim. 30, 6823–6850. Sci. 52, 25–33.
Remer, L.A., Mattoo, S., Levy, R.C., Munchak, L.A., 2013. MODIS 3 km aerosol product: Theil, H., 1992. A Rank-Invariant Method of Linear and Polynomial Regression Analysis.
algorithm and global perspective. Atmos. Measur. Tech. 6, 1829–1844. Henri Theil’s Contributions to Economics and Econometrics, pp. 345–381.
Rémy, S., Kipling, Z., Flemming, J., Boucher, O., Nabat, P., Michou, M., Bozzo, A., Tong, Y., Feng, L., Sun, K., Tang, J., 2020. Assessment of the representativeness of
Ades, M., Huijnen, V., Benedetti, A., Engelen, R., Peuch, V.-H., Morcrette, J.-J., 2019. MODIS aerosol optical depth products at different temporal scales using global
Description and evaluation of the tropospheric aerosol scheme in the European AERONET measurements. Remote Sens. 12.
Centre for Medium-Range Weather Forecasts (ECMWF) Integrated Forecasting Torres, O., Tanskanen, A., Veihelmann, B., Ahn, C., Braak, R., Bhartia, P.K., Veefkind, P.,
System (IFS-AER, cycle 45R1). Geosci. Model Dev. 12, 4627–4659. Levelt, P., 2007. Aerosols and surface UV products from Ozone Monitoring
Richter, A., Burrows, J.P., 2002. Tropospheric NO2 from GOME measurements. Adv. Instrument observations: an overview. J. Geophys. Res. 112. D24S47.
Space Res. 29, 1673–1683. Torres, O., Ahn, C., Chen, Z., 2013. Improvements to the OMI near-UV aerosol algorithm
Rose, A.N., McKee, J.J., Sims, K.M., Bright, E.A., Reith, A.E., Urban, M.L., 2020. using A-train CALIOP and AIRS observations. Atmos. Measur. Tech. 6, 3257–3270.
LandScan 2019. Oak Ridge National Laboratory, Oak Ridge, TN. Ukhov, A., Mostamandi, S., da Silva, A., Flemming, J., Alshehri, Y., Shevchenko, I.,
Salmi, T., Määttä, A., Anttila, P., Ruoho-Airola, T., Amnell, T., Maatta, A., 2002. Stenchikov, G., 2020. Assessment of natural and anthropogenic aerosol air pollution
Detecting Trends of Annual Values of Atmospheric Pollutants by the Mann-Kendall in the Middle East using MERRA-2, CAMS data assimilation products, and high-
Test and Sen's Solpe Estimates the Excel Template Application MAKESENS. Helsinki, resolution WRF-Chem model simulations. Atmos. Chem. Phys. 20, 9281–9310.
Finnish Meteorological Institute, Air Quality Research. ul-Haq, Z., Tariq, S., Ali, M., Mahmood, K., Batool, S.A., Rana, A.D., 2014. A study of
Sayer, A.M., Hsu, N.C., Bettenhausen, C., Jeong, M.-J., 2013. Validation and uncertainty tropospheric NO2 variability over Pakistan using OMI data. Atmos. Pollut. Res. 5,
estimates for MODIS collection 6 “Deep Blue” aerosol data. J. Geophys. Res.-Atmos. 709–720.
118, 7864–7872. ul-Haq, Z., Tariq, S., Rana, A.D., Ali, M., Mahmood, K., Shahid, P., 2015. Satellite remote
Sayer, A.M., Munchak, L.A., Hsu, N.C., Levy, R.C., Bettenhausen, C., Jeong, M.J., 2014. sensing of total ozone column (TOC) over Pakistan and neighbouring regions. Int. J.
MODIS collection 6 aerosol products: comparison between Aqua’s e-deep blue, dark Remote Sens. 36, 1038–1054.
target, and “merged” data sets, and usage recommendations. J. Geophys. Res.- ul-Haq, Z., Tariq, S., Ali, M., 2017. Spatiotemporal assessment of CO2 emissions and its
Atmos. 119, 13965–13989. satellite remote sensing over Pakistan and neighboring regions. J. Atmos. Sol. Terr.
Sayer, A.M., Hsu, N.C., Bettenhausen, C., Jeong, M.J., Meister, G., 2015. Effect of MODIS Phys. 152-153, 11–19.
Terra radiometric calibration improvements on collection 6 deep blue aerosol UNDP, 2019. Sustainable Urbanization. Development Advocate Pakistan.
products: validation and terra/aqua consistency. J. Geophys. Res.-Atmos. 120, Ur Rehman, S.A., Cai, Y., Siyal, Z.A., Mirjat, N.H., Fazal, R., Kashif, S.U.R., 2019. Cleaner
12157–12174. and sustainable energy production in Pakistan: lessons learnt from the Pak-TIMES
Seinfeld, J.H., Pandis, S.N., 1998. Atmospheric Chemistry and Physics, from Air model. Energies 13.
Pollution to Climate Change. John Wiley and Sons, New York. Veefkind, J.P., de Haan, J.F., Brinksma, E.J., Kroon, M., Levelt, P.F., 2006. Total ozone
Sen, P.K., 1968. Estimates of the regression coefficient based on Kendall’s tau. J. Am. from the ozone monitoring instrument (OMI) using the DOAS technique. IEEE Trans.
Stat. Assoc. 63, 1379–1389. Geosci. Remote Sens. 44, 1239–1244.
Shah, M.H., Shaheen, N., Nazir, R., 2012. Assessment of the trace elements level in urban Wang, Y.Q., Zhang, X.Y., Draxler, R.R., 2009. TrajStat: GIS-based software that uses
atmospheric particulate matter and source apportionment in Islamabad, Pakistan. various trajectory statistical analysis methods to identify potential sources from
Atmos. Pollut. Res. 3, 39–45. long-term air pollution measurement data. Environ. Model. Softw. 24, 938–939.
Shen, X., Bilal, M., Qiu, Z., Sun, D., Wang, S., Zhu, W., 2018. Validation of MODIS C6 Wang, Y.Q., Zhang, X.Y., Sun, J.Y., Zhang, X.C., Che, H.Z., Li, Y., 2015. Spatial and
dark target aerosol products at 3 km and 10 km spatial resolutions over the China temporal variations of the concentrations of PM10, PM2.5 and PM1 in China. Atmos.
seas and the eastern Indian Ocean. Remote Sens. 10. Chem. Phys. 15, 13585–13598.
Shi, Y., Zhang, J., Reid, J.S., Hyer, E.J., Hsu, N.C., 2013. Critical evaluation of the MODIS Wang, Q., Sun, L., Wei, J., Yang, Y., Li, R., Liu, Q., Chen, L., 2017. Validation and
Deep Blue aerosol optical depth product for data assimilation over North Africa. accuracy analysis of global MODIS aerosol products over land. Atmosphere 8, 155.
Atmos. Measur. Tech. 6, 949–969. Wang, H., Lu, K., Chen, X., Zhu, Q., Wu, Z., Wu, Y., Sun, K., 2018. Fast particulate nitrate
Shi, Y., Bilal, M., Ho, H.C., Omar, A., 2020. Urbanization and regional air pollution formation via N2O5 uptake aloft in winter in Beijing. Atmos. Chem. Phys. 18,
across south Asian developing countries – a nationwide land use regression for 10483–10495.
ambient PM2.5 assessment in Pakistan. Environ. Pollut. 266. Wang, Y., Yuan, Q., Li, T., Shen, H., Zheng, L., Zhang, L., 2019. Evaluation and
Singh, N., Mhawish, A., Deboudt, K., Singh, R.S., Banerjee, T., 2017. Organic aerosols comparison of MODIS Collection 6.1 aerosol optical depth against AERONET over
over indo-Gangetic plain: sources, distributions and climatic implications. Atmos. regions in China with multifarious underlying surfaces. Atmos. Environ. 200,
Environ. 157, 59–74. 280–301.
Smirnov, A., Holben, B.N., Eck, T.F., Dubovik, O., Slutsker, I., 2000. Cloud-screening and WHO, 2018a. 9 out of 10 People Worldwide Breathe Polluted Air, but More Countries
quality control algorithms for the AERONET database. Remote Sens. Environ. 73, Are Taking Action. Geneva.
337–349. WHO, 2018b. WHO Global Ambient Air Quality Database (Update 2018). Geneva.
Sogacheva, L., de Leeuw, G., Rodriguez, E., Kolmonen, P., Georgoulias, A.K., Xiao, Q., Zhang, H., Choi, M., Li, S., Kondragunta, S., Kim, J., Holben, B., Levy, R.C.,
Alexandri, G., Kourtidis, K., Proestakis, E., Marinou, E., Amiridis, V., Xue, Y., van der Liu, Y., 2016. Evaluation of VIIRS, GOCI, and MODIS Collection 6 AOD retrievals
A, R.J., 2018. Spatial and seasonal variations of aerosols over China from two against ground sunphotometer observations over East Asia. Atmos. Chem. Phys. 16,
decades of multi-satellite observations – part 1: ATSR (1995–2011) and MODIS C6.1 1255–1269.
(2000–2017). Atmos. Chem. Phys. 18, 11389–11407. Xie, Y., Zhang, Y., Xiong, X., Qu, J.J., Che, H., 2011. Validation of MODIS aerosol optical
Song, Z., Fu, D., Zhang, X., Wu, Y., Xia, X., He, J., Han, X., Zhang, R., Che, H., 2018. depth product over China using CARSNET measurements. Atmos. Environ. 45,
Diurnal and seasonal variability of PM2.5 and AOD in North China plain: comparison 5970–5978.
of MERRA-2 products and ground measurements. Atmos. Environ. 191, 70–78. Zhang, R., Sun, X., Shi, A., Huang, Y., Yan, J., Nie, T., Yan, X., Li, X., 2018. Secondary
Stein, A.F., Draxler, R.R., Rolph, G.D., Stunder, B.J.B., Cohen, M.D., Ngan, F., 2015. inorganic aerosols formation during haze episodes at an urban site in Beijing, China.
NOAA’s HYSPLIT atmospheric transport and dispersion modeling system. Bull. Am. Atmos. Environ. 177, 275–282.
Meteorol. Soc. 96, 2059–2077. Zhang, M., Su, B., Bilal, M., Atique, L., Usman, M., Qiu, Z., Ali, M.A., Han, G., 2020. An
investigation of vertically distributed aerosol optical properties over Pakistan using
CALIPSO satellite data. Remote Sens. 12.
28

Remote Sensing of Environment: Sciencedirect

Uploaded by

Copyright:

Available Formats

Remote Sensing of Environment: Sciencedirect

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Remote Sensing of Environment: Sciencedirect

Uploaded by

Copyright:

Available Formats

Remote Sensing of Environment 264 (2021) 112617

Contents lists available at ScienceDirect