1. business and industrial
2. chemicals industry
3. adhesives

Aerosol and Air Quality Research, 15: 682–693, 2015
Copyright © Taiwan Association for Aerosol Research
ISSN: 1680-8584 print / 2071-1409 online
doi: 10.4209/aaqr.2014.04.0088
Discrimination of Aerosol Types and Validation of MODIS Aerosol and Water
Vapour Products Using a Sun Photometer over Central India
Hareef Baba Shaeb Kannemadugu1*, Alappat O. Varghese1, Sivaprasad R. Mukkara1,
Ashok K. Joshi1, Sanjiv V. Moharil2
1
Regional Remote Sensing Centre (central), National Remote Sensing Centre, ISRO-DOS, NBSS-LUP campus, Amaravati
road, Nagpur 440033, India
2
Department of physics, Rashtrasant Tukadoji Maharaj Nagpur University, Amravati road, Nagpur 440033, India
ABSTRACT
Aerosol optical properties were studied using handheld Microtops II sun photometer over Nagpur (79.028°E, 21.125°N)
located in central India, suggests highest Aerosol Optical Depth (AOD) (0.64 ± 0.08) during Pre-Monsoon Season (PMS)
and lowest AOD (0.38 ± 0.06) during Summer Monsoon Season (SMS). Using the scatter plot of AOD500 versus Angstrom
exponent (α), dominating type of aerosol, during the most of the seasons is found to be urban/industrial and biomass
burning (UB). Moderate Resolution Imaging Spectroradiometer (MODIS) active fire data suggests a large amount of
biomass burning in different parts of India during PMS compared to other seasons. Aerosol transport analysis suggests
during PMS, the air masses were originated from the biomass burning regions, desert regions and also from marine
regions. Seasonal variation of AOD550 over India using MODIS data indicates significant AOD (> 0.7) over northwest
during SMS, Indo - Gangetic region during Post Monsoon Season (PoMS) and north east region during winter. Southern
part of India shows relatively less AOD (< 0.45) compared to north India during all the seasons. MODIS derived AOD and
water vapor (NIR), compared against the ground-based observations from sun photometer, indicates an underestimation of
35% lower AOD compared to sun photometer and overestimation of 20% higher water vapor with correlation coefficients
of 0.75 and 0.89 respectively. In view of the sizeable contribution of biomass burning aerosols, we suggest the use of a
more absorbing type of aerosol model for central India, for an accurate retrieval of AOD from MODIS.
Keywords: Aerosol types; central India; Validation; MODIS; Water vapor.
INTRODUCTION
Natural and anthropogenic aerosols are highly variable
in space and time and play a significant role in the earth’s
radiation budget and climate change (Wild, 2009).
Precipitation frequency and rain rate are altered by aerosols
(Li et al., 2011). Black-carbon aerosols absorb solar radiation,
and thus warm the surface while sulfate and organic aerosols
reflect solar radiation and cool the surface. Climate-changemitigation policies are aimed at reducing fossil-fuel blackcarbon emissions, together with the atmospheric ratio of
black carbon to sulfate (Ramana et al., 2010). The high
extinction values observed in the Indian subcontinent and
surrounding regions have a strong impact on regional
climate (Franke et al., 2003). Both the aerosol optical depth
and angstrom exponent depend on the wavelength; hence
*
Corresponding author.
Tel.: +91-712-6621211; Fax: +91-712-6621240
E-mail address: [email protected]
both these parameters can be used for the characterization
of aerosol optical properties (Holben et al., 2001). This lead
to a discrimination of the different aerosol types occurring
in various locations over the globe into four main types:
namely biomass-burning aerosols produced by forest and
grassland fires, urban/industrial aerosols from fossil-fuel
combustion in populated urban/industrial regions, desert
dust blown into the atmosphere by the wind and aerosol of
maritime origin (Kaskaoutis et al., 2007). Efforts have been
made by Indian scientists to carry out annual and inter annual
variability of aerosol parameters using multi-wavelength
radiometer (MWR) since 1980 under the Indian Space
Research Organization (ISRO) Geosphere Biosphere Program
(Murthy et al., 1999). Even so, in view of the strong spatial
and temporal variability of aerosols, the existing groundbased aerosol-monitoring networks in India are not
sufficient to establish aerosol climatology of the whole
country (Tripathi et al, 2005). Aerosol measurements were
reported from several places within the country, but such
data and results are sparse in a dry tropical region in the
central India.
Ground-based observations are important in order to
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
verify the accuracy and validity of satellite-based retrievals.
An extensive validation exercise tests the efficacy of the
retrieval algorithm, conditions under which it works
satisfactorily and cases where further improvement is
needed (Mishra et al., 2008). India has a wide variety of
ecosystems, surface conditions, and emission sources. So it
is important to validate the satellite-based estimates using
ground-based observations for different climatic/ecological
regions throughout the country in order to evaluate and
improve the existing MODIS products. We report the
validation of the MODIS derived aerosol optical depth and
water vapor over the central Indian region.
Long range transport of aerosols plays a significant role
in the observed aerosol loading at a measurement location.
To understand and locate the possible sources, back
trajectory analysis is used. We also used MODIS detected
fire locations in order to understand the contribution of
biomass burning aerosols. The present study is for the first
time over this region focusing on the classification of
aerosol types, validation of MODIS AOD and water vapor
products and the role of aerosol transport.
OBSERVATIONAL SITE AND LOCAL
METEOROLOGY
Central India is bordered with the Great Indian Desert in
the northwest, Indo Gangetic Plain (IGP) in the north and
coastal India in east, west and south. The amazing fact of
the geography of the Nagpur city (21°06′N, 79°03′E), is
683
that it lies at the center of India (Fig. 1) and zero milestone
of India is situated within the city. It has a mean altitude of
310 m above sea level and lies on the Deccan plateau of
Indian peninsula and surrounded by plateaus of Satpura
range. The city of Nagpur enjoys a very dry and semi humid
climate throughout the year except in the monsoon season
(June–September). Dry and hot weather make the climate
scorching almost throughout the pre monsoon (PMS) season
(March, April, May). The maximum temperature remains
more than 42°C and at times; it may reach to 48°C.
Summer Monsoon (SMS) advances in June and extends up
to September. Maximum rainfall occurs during July and
August months. During the post monsoon (PoMS) season
(October and November), the maximum temperature is
about 33°C. Winter season (December, January and February)
of Nagpur is spine chilling with minimum temperatures
around 12°C and at times even dip below that level.
MEASUREMENTS AND DATA ANALYSIS
Data from both the ground-based and satellite-based
instruments are used to study aerosols over the central
Indian region.
Ground Based Measurements
Sunphotmeter:
Model
540
MICROTOPS-II
(microprocessor-based Total Ozone Portable Spectrometer)
sun photometer is a compact, portable and multi-channel
sun photometer, which has been operated, for the first
Fig. 1. Geographical Location of Nagpur in Central India.
684
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
time, at regional remote sensing center (central), Nagpur
(21.125°N, 79.028°E and ~330 m above mean sea level), a
tropical semi-urban station, situated in the central part of
India, to study the characteristics of columnar aerosols and
water vapor and to validate the satellite products.
The physical and operational characteristics of the
instrument are described in the user’s guide (http://www.
solar.com/manuals.htm). The sun photometer measures
solar irradiance in five spectral wave bands from which it
automatically derives AOD. Each has five channels with
peak wavelengths of 440, 500, 675, 870, and 936 nm. The
filters used in all channels have a peak wavelength precision
of ± 1.5 nm and a full width at half maximum (FWHM)
band pass of 10 nm (http://www.solar.com/sunphoto.htm).
Derivation of AOD and water vapor using a sun
photometer has been clearly explained by Ichoku et al.
(2001) and Morys et al. (2001). However; here brief summary
is given.
At 440, 500, 675 and 870 nm wavelengths, AOT is
derived based on the Beer-Lambert-Bouguer law as follows:
Vλ = V0λD–2exp(–τλM),
(1)
where, for each channel (wavelength),
Vλ = the signal measured by the instrument at wavelength λ,
V0λ = the extraterrestrial signal at wavelength λ,
D = Earth-Sun distance in astronomical units at the time of
observation,
τλ = total optical thickness (τλ = τaλ + τRλ+ τO3λ) at
wavelength λ,
τaλ = aerosol optical thickness (AOT) at wavelength λ,
τRλ = Rayleigh (air) optical thickness at wavelength λ,
τO3λ = Ozone optical thickness at wavelength λ,
M = the optical air mass.
The Rayleigh and Ozone optical thickness, τRλ and τO3λ,
are obtained from atmospheric models as below:
τRλ = R4 exp(–h/29.3/273)
(2)
τO3λ = Ozabs × DOBS/1000
(3)
where,
h = altitude of the place of observation in meters,
R4 = 28773.6 × (R2 × (2 + R2) × λ–2)2,
R2 = 10–8 × {8342.13 + 2406030/(130 – λ–2) + 15997/
(38.9 – λ–2)},
λ = wavelength in microns,
Ozabs = Ozone absorption cross section, extracted from a
lookup table based on wavelength,
DOBS = Ozone amount in Dobson units, extracted from a
lookup table based on latitude and date of observation.
Calibration of the MICROTOPS II sun photometer was
performed at the Mauna Loa observatory, Hawaii, a noisefree high-altitude site, by its manufacturer (M/s Solar Light
Control, USA) for reliable results before the measurements
commenced in 2011 at our measurement site. Apart from
this, we continuously monitored the MICROTOPS-II output
at zero air mass (extra-terrestrial constant), which serves as
calibration constant for each channel AOD and columnar
water vapor. Other measurement errors arise due to filter
degradation, temperature effects and poor pointing toward
the sun. When the Microtops is well calibrated and properly
cleaned, its AOT retrievals can be of comparable accuracy
to those of CIMEL sun photometers used in the AERONET
network, with uncertainties in the range of 0.01 to 0.02
(Holben et al., 2001). AERONET stands for Aerosol RObotic
NETwork conceived and maintained by NASA/GSFC
(Holben et al, 1998) and is expanded by collaborators from
other agencies in order to incorporate a large spatial
coverage. It employs CIMEL sun-sky spectral radiometer
which measures the direct sun radiances at eight spectral
channels centered at 340, 380, 440, 500, 670, 870, 940 and
1020 nm.
The data set used here to cover the time period from
January 2011 to December 2011 with total 116 observations.
The ground-based sun photometer measurements are taken
in cloud-free conditions so that there is no cloud patch
covering the line of sight with the sun. For the present
study, sun photometer observations have been selected
from the condition that the time difference between ground
based observation and MODIS overpass time is less than
15 minutes. The data set was used as the ground truth in
the validation of the Terra MODIS AOD550 in this study.
Determination of Angstrom Parameters
Spectral changes in measured AOD values can be
quantified using the angstrom parameters (Angstrom, 1961)
which are estimated using Eq. (4).
τaλ = βλ–α,
(4)
In Eq. (4), τaλ represents the AOD. The angstrom
wavelength exponent (α) indicates the size distribution of
aerosol particles. The turbidity parameter β is a measure of
the aerosol loading and is equal to the AOD measured at
one µm wavelength. The parameters (α, β) were obtained
using the least-squares fit to the plotted logarithm of AOD
versus the logarithm of wavelength.
For validation, AOD from the sun photometer and MODIS
satellite needs to be compared to which AOD from the sun
photometer has been interpolated to a common wavelength
of 550 nm using the following relation,
τ550 = exp[ln(τ500) – ln(550/500)α],
(5)
where α = ln(τ500/τ675)/ln(675/500).
SATELLITE MEASUREMENTS
MODIS
The Moderate Resolution Imaging Spectroradiometer
(MODIS) flies on board the polar orbiting EOS Terra and
Aqua satellite and measure aerosol optical depth and optical
properties on a global scale daily since the year 2000. Terra
and Aqua operate at an altitude of 705 km, with equatorial
crossing time at 10:30 Indian Standard Time (IST) ascending
towards north and at 13:30, IST descending towards south,
respectively. MODIS has 36 bands ranging from 0.4 to
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
14.4 µm wavelengths with three different spatial resolutions
(250, 500 and 1000 m).
In this study, MODIS daily level-3 collection version
005 AOD data at 550 nm averaged at a 1degree latitude/
longitude grid to produce daily MOD08_D3.005 products
from Terra satellite were used. For general climate modeling,
the level 3 data provide a convenient source of data that has
land and ocean measurements at a 1-degree scale combined
into one file. Remer et al. (2005) provided global validation
of Collection 004 (C004) product over both land and ocean
(compared to AERONET) and reported the expected error
bars of AOD values as τpλ = ± 0.05 ± 0.15τpλ over land,
where τpλ is the ground-based AOD value. The updated
C005 algorithm needs to be validated, to account for local
biases. The aerosol properties contained within the lookup
table (LUT) algorithm needs to be updated for as many
sites as possible, in order to make the retrieved AOD more
accurate (Levy et al., 2010).
In order to assess the contribution of forest fires and
agricultural residue burning (which emit black carbon and
other pollutants in to the atmosphere) to the variation of the
aerosol optical depth over India, we used MODIS active fire
data sets downloaded from Fire Information for Resource
Management System (FIRMS) (Davies et al., 2009).
Aerosol Transport: Mapping the Source Locations
The back trajectories are very important to identify the
origin of source regions and the transport pathways of
aerosols reaching the measurement site and also to investigate
the aerosol properties and types (Bian et al., 2011).The back
trajectory analysis provides a three dimensional (latitude,
longitude and altitude) description of the pathways followed
by air mass as a function of time by using National Centre
for Environmental prediction (NCEP) reanalysis or The
Global Data Assimilation System (GDAS) wind as input to
the model. The five days air mass back trajectories at the
height 500 m above the ground level were calculated based on
National Oceanic and Atmospheric Administration (NOAA)
Hybrid Single Particle Lagrangian Integrated Trajectories
(HYSPLIT) model (Draxler and Rolph, 2003).
RESULTS AND DISCUSSION
Seasonal Variation of Aerosol Optical Depth, Angstrom
Parameters
High AOD500 (0.64 ± 0.08) (Fig. 6(c)) is observed during
PMS. High temperature, in association with strong surface
winds, during summer plays an important role in heating
and lifting the loose soil. The incursion of moisture either
from the Bay of Bengal or Arabian Sea and/or operation of
any trigger mechanism create conditions conducive for the
explosive convective phenomenon. This high convective
activity and frequent occurrence of long range transport of
dust from northwestern India lead to increase in AOD
during this season (Flossman et al., 1985).
The monsoon usually advances over central India during
the end of second week or in the third week of June and this
is characterized by severe weather activity like heavy rain,
thundershowers, squally wind, etc. AOD500 values (0.38 ±
685
0.06) are observed to be lower during the monsoon season
due to stronger upper winds, cloud scavenging and rain wash
out processes (Badarinath et al., 2007). The withdrawal of
monsoon is characterized by the reversal of winds from
South West to North East. During the post monsoon, aerosols
build up slowly and possibly undergo hygroscopic growth
in water vapor (RH > 50%) leading to increase in AOD500
(0.5 ± 0.02). The winter season is characterized by dry and
cold weather. In the winter season, AOD500 is less (0.42 ±
0.15) compared to post monsoon season.
Seasonal Variation of Angstrom Exponent (α) and
Turbidity Parameter (β)
Higher values of α (> 1) (Fig. 2(a)) during PoMS and
winter indicate larger abundance of sub micron aerosols (fine
mode aerosols) originating mainly from biomass burning and
fossil-fuel combustion sources. This also shows that the
influence of the dust aerosols is very limited during this
season. The dominating aerosols are in the fine-mode range
formed in-situ by secondary gas-to-particle conversion
process of the precursors (Kaskaoutis et al., 2009).
The values of α are moderate (~0.95) during PMS and
SMS indicating an abundance of coarse mode aerosol
particles and also a reasonable contribution from fine
mode aerosols such as biomass burning. Greater β during
these seasons indicates more aerosol loading. This further
indicates that as the aerosol loading increases, the relative
dominance of fine aerosol decreases. Similar observations
with dominance of fine mode aerosols during winter and
post monsoon (coarse mode aerosols during summer and
monsoon) over other locations namely Gurushikar and
Ahmadabad (Kedia and Ramachandran, 2011) were reported.
Aerosol Type Classification
Better characterization of aerosol properties can be done
using both AOD and α as both are functions of wavelength.
This can be done using scatter plots of AOD versus α from
which different aerosol types can be obtained for a specific
location through determination of physically interpretable
cluster regions of the diagram (Kaskaoutis et al., 2007). This
technique has been used by many authors (e.g., Eck et al.,
1999; Pace et al., 2006; Kaskaoutis et al., 2007; Kalapureddy
et al., 2009; Kaskaoutis et al., 2011; Pathak et al., 2012)
for the discrimination between different aerosol types.
The scatter plot diagram of AOD500 versus α (Fig. 2(b))
has been divided into physically interpretable clusters, in
which each cluster represents the different aerosol type.
For defining the background aerosol type the continental
average model given by Hess et al. (1998) is assumed.
According to this aerosol model, AOD at 550 nm is 0.151
for background aerosols and α in the spectral range 350–
500 nm is 1.11 while in the range 500–800 nm is 1.42. The
average AOD500 over the present study location is ~0.58
while α is found to be 1.11 for 500–870 nm. The threshold
values of continental average (CA) types are taken as
AOD500 < 0.3; α < 0.9. Most of the data points are located in
the cluster region AOD500 > 0.35, α > 1, which corresponds
to the particles of urban/industrial and biomass burning
(UB). Cluster with higher AOD500 (> 0.40) and α < 0.7
686
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
Fig. 2. Optical properties of aerosols (a) Seasonal variation of α and β (b) Scatter plot of Angstrom parameter (α) versus
aerosol optical depth at 500 nm
indicates the presence of Desert Dust (DD). The remaining
points are scattered corresponding to a variety of AOD500
values ranging from ~0.3 to ~0.9 and a wide range of α
(~0.3 to ~1.5). These are categorized as the undetermined
or mixed type (MT) aerosols (Cachorro et al., 2001). This
mixed or undetermined aerosol type is formed possibly by
the mixture of anthropogenic and natural aerosols.
The seasonal distribution of the percentage contributions
of the different aerosol types calculated using the scatter
plot diagram of AOD500 versus α is presented in Fig. 3.
During the post monsoon and winter, dominant aerosol
type is UB followed by the undetermined or mixed type
(MT). The presence of small amounts of Desert Dust (DD)
and continental average (CA) type aerosols in PoMS and
winter respectively is observed. During winter season, the
anthropogenic emissions (fossil-fuel combustion, black
carbon emissions) are found to be large over India
(Ramachandran and Rajesh, 2007; Pathak et al., 2010).
During PMS, UB and DD aerosol, types dominate and
during SMS dominant aerosol type is MT followed by UB
and DD respectively.
MODIS detected fire locations: In India, 55% of the
total forest cover is prone to fires annually (Gubbi, 2003),
which are mainly attributed to anthropogenic factors, like
slash and burn agricultural practices, controlled burning,
deforestation, fire-wood burning and others. Vadrevu et al.
(2013) done a detailed study of these forest fires in India
they attributed, fires in Punjab region and Indo-Ganges
region, to agricultural residue burning and burning of closed
broad leaf deciduous forests. In southern India, the fires
are attributed to slash and burn agriculture, locally known
as “Podu”. In central India (in and around measurement
location) are attributed to biomass burning of crop residues,
mostly Soybean and Wheat and forest burning due to
intentional fires of tropical dry deciduous forests. The forests
are burnt for clearing of land for agriculture, facilitating
growth of grass for cattle grazing, in addition to management
fires by the local forest department. In the northeast region,
large amount of fires in the region is attributed to slash and
burn agriculture also known as Jhum, practiced by the
local people.
Fig. 4 shows MODIS sensor detected fire locations for
different seasons. During PMS, there is a large amount of
biomass burning in distinct parts of India. During other
seasons, the biomass burning is observed to be low except
in the Punjab region during post monsoon, south west and
north east regions during winter.
Role of Aerosol transport: Source apportionment: Fig. 4
shows the locations of sources of air masses reaching at
the measurement site for different seasons namely, PMS,
SMS, PoMS and winter. During PMS, the air masses were
originated from the biomass burning regions (see Fig. 4),
desert regions and also from marine regions. During SMS,
because of the sustained south westerly flow of the monsoon
winds, the air masses were mostly of marine origin. During
the post monsoon season, the dominant air masses were
originated from north India, including transport of air masses
from biomass burning, Punjab region. During winter, the
origins of air masses were located in eastern India and IGP
region.
Validation of MODIS Aerosol and Water Vapor Products
Seasonal variability of AOD over India: Fig. 5 shows
seasonal average MODIS Terra AOD550 over the Indian
subcontinent and surrounding regions. Seasonal variations
in AOD are found to be significant (AOD > 0.7) over
northwest during SMS, IGP during PoMS and North east
during winter. Southern part of India shows relatively less
AOD (< 0.45) compared to north India during all the
seasons.Consistently high AOD in IGP region during all the
seasons and inability to capture significant AOD seasonal
changes in south India, indicate that there may be a
significant bias in MODIS retrieval algorithm. This is due
to adjusting of its algorithm based on AERONET station
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
687
Fig. 3. Fraction pies of each aerosol type in different seasons.
(IIT Kanpur: 26.28°N, 80.24°E) data and in this process,
the AOD’s retrieved over other parts of India may be
significantly under/over estimated. It is this point we would
like to explore further by validating AOD over Nagpur
located in central India.
Validation of MODIS Aerosol Optical Thickness
The purpose of validating any satellite-based geophysical
parameter is to understand the errors involved in retrieval
of such parameters and later correct the retrieval algorithm
accordingly. Moreover, increase in the number of validated
regions will lead to better comparison of the parameters on a
global scale.For this purpose, detailed validation of MODIS
AOD products of different versions with distinct spatial
resolutions by using the ground-based multi wavelength
radiometer and MICROTOPS sun photometer has been
performed by several authors over the Indian subcontinent
(e.g., Tripathi et al, 2005; Prasad et al., 2007; Mishra et
al., 2008; Vinoj et al., 2008; Guleria et al., 2012). The
studies found MODIS overestimating the AOD values during
the summer and underestimating during winter. The present
study attempts to validate the latest MODIS aerosol product
version C005 over the central Indian region where there is
no validation exercise done so far.
It is seen from the Fig. 6 that the MODIS AOD values
were relatively lower than the sun photometer AOD values.
The scatter plot (Fig. 6(a)) of sun photometer AOD versus
MODIS AOD showed good agreement for daily observations.
The linear regression line between AOD derived from
MODIS and sun photometer was found to be
MODIS AOD = 0.65 (sun photometer AOD) − 0.03,
(R = 0.75)
(6)
A high correlation of 0.75 observed indicates that the
MODIS can capture the seasonal variability well, and a
slope of 0.65 implies an underestimation of 35% lower
AOD compared to sun photometer. The absolute errors in
retrieval of MODIS AOD show a linear relationship with
its AOD and were found to be given as follows:
Absolute error = −0.13 (MODIS AOD) + 0.26,
(7)
Fig. 6(d) shows that the absolute error in estimation of
AOD by MODIS decreases for an increase in aerosol
loading. We see that the MODIS–sun photometer AOD
correlation line has a small negative offset and a slope less
than one (Levy et al., 2005; Tripathi et al., 2005). The yaxis offsets imply errors induced by assuming inappropriate
surface reflectance. On the other hand, the less-than-one
slope implies errors in the aerosol models (Levy et al., 2005).
A similar deviation of the slope from unity was observed
by several authors (e.g., Tripathi et al., 2005; Mishra et al.,
2008; Gloria et al., 2012), which they attributed to error in
aerosol model assumptions. Below we discuss possible
error sources.
To understand the results, linear regression lines were
fitted season wise as shown in Table 1. In the MODIS
688
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
Fig. 4. MODIS active fire data (red color symbols) and locations of sources of air masses (filled yellow color circles)
reaching at the measurement site and overlaid on ESRI (http:/www.esri.com) world image layer for different seasons PMS,
SMS, PoMS and winter
AOD retrieval algorithm, by default neutral aerosol model
(Single Scattering Albedo (SSA) ~0.9) was set for a major
part of Asia (Remer et al., 2005; Levy et al., 2007) for
different seasons throughout a year. Absorbing (SSA ~0.85)
or non-absorbing (SSA ~0.95) models were applied in
other parts around the world. This is based on the aerosol
types observed in AERONET sites located at different parts
around the world and based on the condition that If either
the non-absorbing or the absorbing aerosol occupied more
than 40% of the pie, and the other occupied less than 20%,
then the site was designated as the dominant aerosol type.
In India, there is only one AERONET site (IIT Kanpur:
26.28°N, 80.24°E) located within the IGP region and
aerosol types derived there is used in the retrieval of AOD
for other locations.
During PMS main aerosol types observed over Nagpur
location, were UB and DD, but MODIS algorithm assumes
neutral aerosol while there is a good percentage of UB is
present, which is of absorbing type of aerosol. The absolute
error between AOD measured by the sun photometer to
that of MODIS retrieved AOD is maximum (0.29) for this
season. A good number of MODIS fire locations over
central India and back trajectory analysis also indicate that
there is a transport from such locations. This will lead to
underestimation of AOD, as for absorbing aerosols if the
algorithm considers the scattering nature of aerosols, it will
wrongly assign a smaller value of AOD to match calculated
radiance with observed radiance leading to underestimation
of the retrieved AOD. The SSA for black carbon aerosols
due to biomass burning is significantly lower than that of
dust aerosols (Myhre et al., 2005). This may be the reason,
while other authors reported overestimation of MODIS
AOD compared to ground based AOD, over this region, we
had observed the underestimation because of the significant
amount of black carbon.
Similarly, during PoMS and winter seasons also presence
of substantial amounts of UB leads to underestimation,
while the variation in percentage of underestimation may be
due to the presence of another type of aerosol such as mixed
type of aerosol. During SMS, it is difficult to interpret
because of the dominating presence of mixed type of
aerosol and also presence of cloud contamination and less
no of data points.
It is observed from Table 1 that the PoMS and winter
season were having lower negative offset compared to PMS
and SMS, while negative offset implies the overestimation
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
689
Fig. 5. Seasonally averaged AOD550 from MODIS-Terra for Indian subcontinent
of surface reflectance. Again this is because the surface type
assumed, for India is based upon the AERONET site located
at Kanpur. MODIS algorithm assumes that many of the
AERONET sites in the tropics are in savanna or grassland
regions, where the landscape is not as green, and hence the
reds to SWIR ratios are also lower. Significant y-offset
observed during different seasons also indicate that there is
an inadequacy in surface reflectance assumption for the
heterogeneous surfaces consisting of several land features
(Water bodies, settlements, vegetated surface etc.). So Land
surface features also need to be included in the pixel level
(may be by using high resolution land imagery products),
in order to reduce the offset and seasonal dependence of
surface type needs to be taken into the retrieval model.
Overall, the validation results show that the MODIS
AOD retrievals do not represent, accurately, true
observations in central India, and thus cannot be well
applied there. This can be attributed to a complexity of
surface conditions and aerosol types and seasonal changes
of surface reflectance and aerosol models over different
ecological and geographic regions. Hence we suggest a
higher absorbing type of model and also include seasonally
changing land use/land cover features in central India for
an accurate retrieval.
Seasonal variation and Validation of columnar water
vapour: Columnar water vapor (CWC) amount measured
using a sun photometer over this region is typically in the
0.4–4 cm range. A seasonal variation of CWC values
during the year 2011 is shown in Fig. 7(c). There exists a
well-defined seasonal variation in CWC, with the
maximum value during the monsoon months and minimum
during winter months. Similar variations in CWC have
been observed from other locations in India (e.g., Moorthy
et al., 1999; Balakrishnaiah et al., 2011).The validation of
gridded products (MODIS) is important as they are used
for assimilation in numerical weather prediction and global
climate models (Prasad et al., 2009). For this purpose,
detailed validation of MODIS water vapor product is
attempted.
It is seen from the Fig. 7 that the MODIS water vapor
(WV) values were relatively higher than the water vapor
values measured by the sun photometer. The scatter plot
(see Fig. 7(a)) of sun photometer WV versus MODIS WV
showed good agreement for daily observations. The linear
regression line between WV derived from MODIS and sun
photometer was found to be:
MODIS WV (NIR) = 1.2 (sun photometer WV) − 0.3, (8)
(R = 0. 89)
The absolute errors in retrieval of MODIS WV show a
linear relationship with its WV and were found to be given
as follows:
Absolute error = 0.3 (MODIS WV (NIR)) − 0.1,
(9)
Fig. 7(d) shows that the absolute error in estimation of
WV by MODIS increases with an increase in water vapour
content. In the case of SMS this may be the reason for high
absolute error because of high rainfall.
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
690
Fig. 6. (a) A scatter plot between AOD derived from MODIS and Sun photometer: linear regression line and 1:1 line is
shown. Variations in (b) monthly and (c) seasonal-averaged AOD550 retrieved from MODIS and sun photometer; the error
bars in panels (b) and (c) represent the standard deviation of AOD. (d) A scatter plot of the absolute error in MODIS AOD
retrieval and MODIS AOD.
Table 1. Linear regression parameters for different seasons.
Season
PMS
SMS
PoMS
Winter
Linear regression line
MODIS AOD = 0.76 (Sun photometer AOD) − 0.13
MODIS AOD = 1.45 (Sun photometer AOD) − 0.35
MODIS AOD = 0.79 (Sun photometer AOD) − 0.06
MODIS AOD = 0.67 (Sun photometer AOD) − 0.03
Over this location, it is seen that the turbidity parameter
is maximum during PMS and SMS seasons and minimum
during PoMS and winter seasons. This may be the reason for
observed errors. This is in accordance with the observation
that under hazy conditions (with visibilities less than 10
km) or when the surface reflectance near 1 µm is small
(less than about 0.1), errors can be 10% or slightly greater
in our retrieved water vapor values using the atmospheric
transmittance model (Gao and Kaufman, 2003). Their
result was according to the validation performed using the
microwave radiometer but in our case, we used a sun
photometer for the validation. The discrepancy during the
SMS may be due to the least number of data points.
R
0.62
0.85
0.77
0.88
No. of samples
38
7
20
37
CONCLUSIONS
This paper presents seasonal variability of Aerosol optical
depth, Angstrom parameters, aerosol types and water vapour
over Nagpur, central India. The role of forest and biomass
burning fires and transport of air masses was also analyzed.
Seasonal variability of AOD over India is analyzed using
MODIS C005 Aerosol data products and validation of
MODIS AOD and water vapour is performed.
The main conclusions of the study are summarized as
follows:
1. Four different aerosol types have been classified from
the relation between AOD500 and α, applying appropriate
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
691
Fig. 7. (a) A scatter plot between columnar water derived from MODIS and Sun photometer: Regression line and 1:1 line
is shown. Variations in (b) monthly and (c) seasonal-averaged AOD retrieved from MODIS and MWR; the error bars in
panels (b) and (c) represent the standard deviation of AOD. (d) A scatter plot of the absolute error in MODIS AOD
retrieval and MODIS AOD.
threshold values. The percentage contribution of each
type varies seasonally. UB aerosol type dominates all the
seasons except summer monsoon season. The significant
contribution of desert dust during Pre monsoon and
summer monsoon is present.
2. MODIS active fire data suggest there is a large amount
of biomass burning during the pre monsoon season,
which may be the reason for the observed high contribution
of UB aerosol type observed during this season. Air
mass back trajectory analysis suggests that the sources
were located in biomass burning, desert and marine
regions. During the post monsoon season, the dominant
air masses were originated from north India and in
winter, the origins of air masses were located in eastern
India and IGP region.
3. Seasonal variations in MODIS retrieved AOD550 over
India suggests high AOD over northwest during SMS,
IGP region during PoMS and North east during winter.
Southern part of India shows relatively less AOD
compared to north India during all seasons.
4. Validation of MODIS TERRA retrieved AOD550 with
Sun photometer suggests underestimation by MODIS,
which has been attributed to errors in aerosol model and
surface reflectance assumptions. Instead of the neutral
aerosol model assumed in the MODIS retrieval algorithm,
we recommend more absorbing type of model in view of
the sizeable contribution of biomass burning aerosols in
the central Indian region and India at large.
5. There exists a well-defined seasonal variation in the
columnar water vapor, with the maximum value during the
monsoon months and minimum during the winter months.
Validation of MODIS TERRA retrieved water vapor
(NIR) with Sun photometer suggests overestimation by
MODIS, which has been attributed to errors due to
turbidity or haze in the atmosphere.
Aside from the above conclusions, it is confirmed that
the MODIS aerosol and water vapor products cannot be
used all over the globe with same accuracy .Users should
validate the products before the use, for reliable conclusions.
This caution is applicable for MODIS data derived air
pollution studies etc also. This also points out at the need
for new kind of technique such as polarizing airborne
692
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
instrument for spatial studies of the Aerosols in India.
ACKNOWLEDGMENTS
We thank Dr. V.K. Dhadwal, Dr. J.R. Sharma and Dr.
C.B.S. Dutt, NRSC for their support and encouragement
throughout this study. We acknowledge NASA for
providing MODIS and aerosol data and the NOAA Air
Resources Laboratory for the provision of the HYSPLIT
transport and dispersion model. We thank FIRMS (Fire
Information for Resource Management System) team for
making data available through the website. Thanks are also
due to the Editor and the anonymous reviewer for their
critical comments which helped to improve the quality of
this manuscript.
REFERENCES
Ångström, A. (1961). Techniques of Determining the
Turbidity of the Atmosphere. Tellus13: 214.
Badarinath, K.V.S., Kharol, S.K., Kaskaoutis, D.G. and
Kambezidis, H.D. (2007). Dust Storm over Indian Region
and Its Impact on the Ground Reaching Solar Radiation
Case Study Using Multi-satellite Data and Ground
Measurements. Sci. Total Environ. 384: 316–332.
Balakrishnaiah, G., Kumar, K.R., Reddy, B.S.K., Gopal,
K.R., Reddy, R.R., Reddy, L.S.S., Ahammed, Y.N.,
Narasimhulu, K., Moorthy, K.K. and Babu, S.S. (2011).
Analysis of Optical Properties of Atmospheric Aerosols
Inferred from Spectral AODs and Angstrom Wavelength
Exponent. Atmos. Environ. 45: 1275–1285.
Bian, H., Tie, X., Cao, J., Ying, Z. Han, S. and Xue, Y.
(2011). Analysis of a Severe Dust Storm Event over China:
Application of the WRF-Dust Model. Aerosol Air Qual.
Res. 11: 419–428.
Cachorro, V.E., Vergaz, R. and de Frutos, A.M. (2001). A
Quantitative Comparison of α-Å Turbidity Parameter
Retrieved in Different Spectral Ranges Based on
Spectroradiometer Solar Radiation Measurements.
Atmos. Environ. 35: 5117–5124.
Davies, D.K., Ilavajhala, S., Wong, M.M. and Justice, C.O.
(2009). Fire Information for Resource Management
System: Archiving and Distributing MODIS Active Fire
Data. IEEE Trans. Geosci. Remote Sens. 47: 72–79.
Draxler, R.R. and Rolph, G.D. (2003). HYSPLIT (Hybrid
Single Particle Lagrangian Integrated Trajectory) Model,
Report, Air Resource Laboratory, NOAA, Silver, Spring,
Md. (Available at: http://www.arl.noaa.gov/ready/hyspl
it4.html).
Eck, T.F., Holben, B.N., Reid, J.S., Dubovik, O., Smirnov,
A., O'Neill, N.T., Slutsker, I. and Kinne, S. (1999). The
Wavelength Dependence of the Optical Depth of Biomass
Burning, Urban, and Desert Dust Aerosols. J. Geophys.
Res. 104: 31333–31349.
Flossmann, A.I., Hall, W.D. and Pruppacher, H.R. (1985).
A Theoretical Study of the Wet Removal of Atmospheric
pollutants. 1. The Redistribution of Aerosol-particles
Captured through Nucleation and Impaction Scavenging
by Growing Cloud Drops. J. Atmos. Sci. 42: 583–606.
Franke, K., Ansmann, A., Muller, D., Althausen, D.,
Venkataraman, C., Reddy, M.S., Wagner, F. and Scheele,
R. (2003). Optical Properties of the Indo-Asian Haze
Layer over the Tropical Indian Ocean. J. Geophys. Res.
108: 1–17.
Gao, B.C. and Kaufman, Y.J. (2003). Water Vapor Retrievals
Using Moderate Resolution Imaging Spectroradiometer
(MODIS) Near-infrared Channels. J. Geophys. Res. 108:
4389.
Gubbi, S. (2003). Fire, Fire Burning Bright! Deccan
Herald. (Accessed January 2004). (Available Online at
http://wildlifefirst.info/images/wordfiles/fire.doc).
Guleria, R.P., Kuniyal, J.C., Rawat, P.S., Thakur, H.K.,
Sharma, M., Sharma, N.L., Dhyani, P.P. and Singh, M.
(2012). Validation of MODIS Retrieval Aerosol Optical
Depth and an Investigation of Aerosol Transport over
Mohal in North Western Indian Himalaya. Int. J. Remote
Sens. 33: 5379–5401.
Hess, M., Koepke, P. and Schult, I. (1998). Optical Properties
of Aerosols and Clouds: The Software Package OPAC.
Bull. Am. Meteorol. Soc. 79: 831–844.
Holben, B.N., Eck, T.F., Slutsker, I.D., Tanré, D., Buis, J.P.,
Setzer, A., Vermote, E., Reagan, J.A., Kaufman, Y.J.,
Nakajima, T., Lavenu, F., Jankowiak, I. and Smirnov,
A. (1998). AERONET-A Federated Instrument Network
and Data Archive for Aerosol Characterization. Remote
Sens. Environ. 66: 1–16.
Holben, B.N., Tanre, D., Smirnov, A., Eck, T.F., Slutsker,
I., Abuhassan, N., Newcomb, W.W., Schafer, J.S.,
Chatenet, B., Lavenu, F., Kaufman, Y.J., Castle, J.V.,
Setzer, A., Markham, B., Clark, D., Frouin, R., Halthore,
R., Karneli, A., O'Neill, N. T., Pietras, C., Pinker, R.T.,
Voss, K. and Zibordi, G. (2001). An Emerging Groundbased Aerosol Climatology: Aerosol Optical Depth from
AERONET. J. Geophys. Res. 106: 12067–12097.
Ichoku, C., Levy, R., Kaufman, Y.J., Remer, L.A., Li,
R.R., Martins, V.J., Holben, B.N., Abuhassan, N., Slutsker,
I., Eck, T.F. and Pietras, C. (2001). Analysis of the
Performance Characteristics of the Five-channel Microtops
II Sun Photometer for Measuring Aerosol Optical
Thickness and Precipitable Water Vapor. J. Geophys.
Res. 106: 14573–14582.
Kalapureddy, M.C.R., Kaskaoutis, D.G., Raj, P.E., Devara,
P.C.S., Kambezidis, H.D., Kosmopoulos, P.G. and Nastos,
P.T. (2009). Identification of Aerosol Type over the
Arabian Sea in the Pre-monsoon Season during the
Integrated Campaign for Aerosols, Gases and Radiation
Budget (ICARB). J. Geophys. Res. 114, doi: 10.1029/
2009JD011826.
Kaskaoutis, D.G., Kambezidis, H.D., Hatzianastassiou, N.,
Kosmopoulos, P.G. and Badarinath, K.V.S. (2007).
Aerosol Climatology: On the Discrimination of Aerosol
Types over four AERONET Sites. Atmos. Chem. Phys.
Discuss. 7: 6357–6411.
Kaskaoutis, D.G., Kambezidis, H.D., Hatzianastassiou, N.,
Kosmopoulos, P.G. and Badarinath, K.V.S. (2007).
Aerosol Climatology: Dependence of the Angstrom
Exponent on Wavelength over Four AERONET Sites.
Atmos. Chem. Phys. Discuss. 7: 7347–7397.
Kannemadugu et al., Aerosol and Air Quality Research, 15: 682–693, 2015
Kaskaoutis, D.G., Badarinath, K.V.S., Kharol, S.K., Sharma,
A.R. and Kambezidis, H.D. (2009). Variations in the
Aerosol Optical Properties and Types over the Tropical
Urban Site of Hyderabad, India. J. Geophys. Res. 114:
1–20.
Kaskaoutis, D.G., Kharol, S.K., Sinha, P.R., Singh, R.P.,
Kambezidis, H.D., Sharma, A.R. and Badarinath, K.V.S.
(2011). Extremely Large Anthropogenic-aerosol
Contribution to Total Aerosol Load over the Bay of
Bengal during Winter Season. Atmos. Chem. Phys. 11:
7097–7117.
Kedia, S. and Ramachandran, S. (2011). Seasonal Variations
in Aerosol Characteristics over an Urban Location and a
Remote Site in Western India. Atmos. Environ. 45:
2120–2128
Levy, R.C., Remer, L.A., Martins, J.V., Kaufman, Y.J.,
Plana-Fattori, A., Redemann, J. and Wenny, B. (2005).
Evaluation of the MODIS Aerosol Retrievals over Ocean
and Land during CLAMS. J. Atmos. Sci. 62: 974–992.
Levy, R.C., Remer, L.A. and Dubovik, O. (2007). Global
Aerosol Optical Properties and Application to Moderate
Resolution Imaging Spectroradiometer Aerosol Retrieval
over Land. J. Geophys. Res. 112: 1–15.
Levy, R.C., Remer, L.A., Kleidman, R.G., Mattoo, S., Ichoku,
C., Kahn, R. and Eck, T.F. (2010). Global Evaluation of
the Collection 5 MODIS Dark-target Aerosol Products
over Land. Atmos. Chem. Phys. 10: 10399–10420.
Li, Z., Niu, F., Fan, J., Liu, Y., Rosenfeld, D. and Ding, Y.
(2011). Long-term Impacts of Aerosols on the Vertical
Development of Clouds and Precipitation. Nat. Geosci.
4: 888–894.
Misra, A., Jayaraman, A. and Ganguly, D. (2008). Validation
of MODIS Derived Aerosol Optical Depth over Western
India. J. Geophys. Res. 113: D04203.
Moorthy, K.K., Nair, P.R., Prasad, B.S.N., Muralikrishnan,
N., Gayathri, H.B., Murthy, B.N., Niranjan, K., Ramesh
Babu, Y.Y., Satyanarayana, G.V., Agashe, V.V., Aher,
G.R., Singh, R. and Srivastava, B.N. (1993). Results from
the MWR Network of IMAP. Indian J. Radio Space
Phys. 22: 243–258.
Moorthy, K.K., Niranjan, K., Narasimhamurthy, B., Agashe,
V.V. and Murthy, B.V.K. (1999). Aerosol Climatology
over India: 1-ISRO GBP MWR Network and Database.
Indian Space Research Organization, Bangalore, India.
Morys, M., Mims III, F.M., Hagerup, S., Anderson, S.E.,
Baker, A., Kia, J. and Walkup, T. (2001). Design,
Calibration, and Performance of Microtops II handheld
Ozone Monitor and Sun Photometer. J. Geophys. Res.
106: 14573–14582.
Myhre, G., Govaerts, Y., Haywood, J.M. and Berntsen, T.
K. (2005). Radiative Effect of Surface Albedo Change
from Biomass Burning. Geophys. Res. Lett. 32: 20812.
Pace, G., DI Sarra, A., Meloni, D., Piacentino, S. and
Chamard, P. (2006). Aerosol Optical Properties at
Lampedusa (Central Mediterranean). 1. Influence of
Transport and Identification of Different Aerosol Types.
693
Atmos. Chem. Phys. 6: 697–713.
Pathak, B., Kalita, G., Bhuyan, K., Bhuyan, P.K. and
Moorthy, K.K. (2010) Aerosol Temporal Characteristics
and Its Impact on Shortwave Radiative Forcing at a
Location in the Northeast of India. J. Geophys. Res. 115:
1–14.
Pathak, B., Bhuyan, P.K., Gogoi, M. and Bhuyan, K.
(2012). Seasonal Heterogeneity in Aerosol Types over
Dibrugarh-North-Eastern India. Atmos. Environ. 47: 307–
315.
Prasad, A.K., Singh, S., Chauhan, S.S., Srivastava, M.K.,
Singh, R.P. and Singh, R. (2007) Aerosol Radiative
Forcing over the Indo-Gangetic Plains during Major Dust
Storms. Atmos. Environ. 41: 6289–6301.
Prasad, A.K. and Singh, R.P. (2009). Validation of
MODIS Terra, AIRS, NCEP/DOE AMIP-II Reanalysis2, and AERONET Sun Photometer Derived Integrated
Precipitable Water Vapor Using Ground-based GPS
Receivers over India. J. Geophys. Res. 114: D05107.
Ramachandran, S. and Rajesh, T.A. (2007). Black Carbon
Aerosol Mass Concentrations over Ahmedabad, an Urban
Location in Western India: Comparison with Urban
Sites in Asia, Europe, Canada, and the United States. J.
Geophys. Res. 112: 1–19.
Ramana, M.V., Ramanathan, V., Feng, Y., Yoon, SC., Kim,
SW., Carmichael, G.R. and Schauer, J.J. (2010). Warming
Influenced by the Ratio of Black Carbon to Sulfate and
the Black-carbon Source. Nat. Geosci. 3: 542–545.
Remer, L., Kaufman, Y.J., Tanre, D., Mattoo, S., Chu, D.A.,
Martins, J., Li, R.R., Ichoku, C., Levy, R.C., Kleidman,
R.G., Eck, T.F., Vermote, E. and Holben, B.N. (2005).
The MODIS Aerosol Algorithm, Products, and Validation.
J. Atmos. Sci. 62: 947–973.
Satheesh, S.K., Moorthy, K.K., Babu, S.S., Vinoj, V. and
Dutt, C.B.S. (2008). Climate Implications of Large
Warming by Elevated Aerosol over India. Geophys. Res.
Lett. 35: L19809.
Tripathi, S.N., Dey, S., Chandel, A., Srivastava, S., Singh,
R.P. and Holben, B.N. (2005). Comparison of MODIS
and AERONET Derived Aerosol Optical Depth over the
Ganga Basin, India. Ann. Geophys. 23: 1093–1101.
Vadrevu, K.P., Csiszar, I., Ellicott, E., Giglio, L., Badarinath,
K.V.S., Vermote, E. and Justice, C. (2013). Hotspot
Analysis of Vegetation Fires and Intensity in the Indian
Region. IEEE J. Sel. Topics Appl. Earth Observ. 6: 224–
238.
Vinoj, V., Satheesh, S.K. and Moorthy, K.K. (2008).
Aerosol Characteristics at a Remote Island: Minicoy in
Southern Arabian Sea. J. Earth Syst. Sci. 117: 389–397.
Wild, M. (2009). Global Dimming and Brightening: A
Review. J. Geophys. Res. 114: 1–31.
Received for review, May 29, 2014
Revised, August 28, 2014
Accepted, September 26, 2014