Variation of column-integrated aerosol properties in a Chinese urban region

Abstract

Thirty-three months of aerosol data in Beijing are presented in this paper. Aerosol optical thickness (AOT) increases from January to June and then decreases gradually. However, airborne particulate matter with diameter less than 10 μm (PM10) concentration exhibits higher values in winter and spring and lower concentration in summer. For the same PM10 concentration, AOT in summer is approximately two, three, and four times that in autumn, winter, and spring, respectively. AOT increases persistently during daytime, and the diurnal variation varies from about 15% in summer to about 45% in winter. The seasonal and diurnal variation of AOT is quite different from that of surface particle concentration. This is partly attributed to the variation of atmospheric mixing layer height. Aerosol volume concentrations increase with AOT by nearly identical magnitude for fine and coarse mode except in spring. The volume concentration of coarse mode in spring increases by a magnitude of more than two times that derived in remaining seasons. Aerosol fine mode radius increases with AOT, whereas coarse mode radius keeps relatively invariable with AOT. Mean aerosol single-scattering albedo at 440 nm is about 0.90 and decreases slightly with wavelength. Aerosol single-scattering albedos increase and their spectral dependence reverses during dust periods. Aerosol size and absorption in Beijing are close to results derived in Mexico City and Kanpur, but they are quite different from those in Maryland and Paris. Therefore different urban aerosol models should be created and used in satellite remote sensing in different urban regions.

Full text

Variation of column-integrated aerosol
properties in a Chinese urban region
X. A. Xia,
1
H. B. Chen,
1
P. C. Wang,
1
W. X. Zhang,
1
P. Goloub,
2
B. Chatenet,
2
T. F. Eck,
3
and B. N. Holben
3
Received 11 May 2005; revised 15 August 2005; accepted 28 October 2005; published 7 March 2006.
[1] Thirty-three months of aerosol data in Beijing are presented in this paper. Aerosol
optical thickness (AOT) increases from January to June and then decreases gradually.
However, airborne particulate matter with diameter less than 10 mm (PM
10
) concentration
exhibits higher values in winter and spring and lower concentration in summer. For
the same PM
10
concentration, AOT in summer is approximately two, three, and four times
that in autumn, winter, and spring, respectively. AOT increases persistently during
daytime, and the diurnal variation varies from about 15% in summer to about 45% in
winter. The seasonal and diurnal variation of AOT is quite different from that of surface
particle concentration. This is partly attributed to the variation of atmospheric mixing
layer height. Aerosol volume concentrations increase with AOT by nearly identical
magnitude for fine and coarse mode except in spring. The volume concentration of coarse
mode in spring increases by a magnitude of more than two times that derived in remaining
seasons. Aerosol fine mode radius increases with AOT, whereas coarse mode radius
keeps relatively invariable with AOT. Mean aerosol single-scattering albedo at 440 nm is
about 0.90 and decreases slightly with wavelength. Aerosol single-scattering albedos
increase and their spectral dependence reverses during dust periods. Aerosol size and
absorption in Beijing are close to results derived in Mexico City and Kanpur, but they are
quite different from those in Maryland and Paris. Therefore different urban aerosol models
should be created and used in satellite remote sensing in different urban regions.
Citation: Xia, X. A., H. B. Chen, P. C. Wang, W. X. Zhang, P. Goloub, B. Chatenet, T. F. Eck, and B. N. Holben (2006), Variation of
column-integrated aerosol properties in a Chinese urban region, J. Geophys. Res., 111, D05204, doi:10.1029/2005JD006203.
1. Introduction
[2] It has been recognized that aerosol particles influence
the Earth’s radiative balance directly by backscattering and
absorption of shortwave (solar) radiation and indirectly by
influencing cloud properties and lifetime [Charlson et al.,
1992]. The increase of aerosol loading is likely to account
partly for a notable decrease of sunshine duration and
downwelling surface solar irradiance [Kaiser and Qian,
2002; Xia, 2004]. Regional temperature change in Sichuan
Basin and precipitation pattern change from the mid- 1970s
were suggested to be related to the heavy aerosol loading and
aerosol strong absorption in China [Li et al., 1995; Xu, 2001;
Qian and Giorgi, 2000; Menon et al., 2002]. However, our
knowledge of aerosol effects on climate and environment is
still limited. One of the reasons is that measurements of
aerosol parameters on a global/regional scale are not com-
plete, for example, in China. Therefore long-term, detailed
global measurements from satellites and well-distributed
ground networks are urgently required in order to character-
ize aerosol properties and to study their effects on climate
and environment [Kaufman et al., 2002; Holben et al., 1998].
[
3] In China, rapid economic growth and population
expansion in the last 20 years led to a significant increase
of aerosol optical thickness (AOT) over much of China.
AOT at 750 nm observed from 46 stations in China increased
from 0.38 in 1960 to 0.47 in 1990 [Luo et al., 2001].
Specifically, Chinese cities experience high airborne particle
concentrations because of airborne dust and primary par-
ticles emitted from coal and biomass combustion, motor
vehicle exhaust, as well as secondary sulfates formed from
the sulfur dioxide by atmospheric chemical reaction. Heavy
aerosol loading in China has been reported on the basis of
observations by ground-based instruments and satellites. Xia
et al. [2005] studied aerosol physical and radiative properties
and their spatial and temporal variation over north China on
the basis of ground-based remote sensing data in spring
2001. The transportation of dust and industrial pollution to
the downwind regions has been discussed by Eck et al.
[2005] using Aerosol Robotic Network (AERONET ) data
and by Husar et al. [2001] using satellite and ground-based
data. Bergin et al. [2001] measured aerosol radiative prop-
erties and chemical compositions in Beijing during July
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 111, D05204, doi:10.1029/2005JD006203, 2006
1
Institute of Atmospheric Physics, Chinese Academy of Sciences,
Beijing, China.
2
Laboratoire d’Optique Atmosphe´rique, Universite´ des Sciences et
Technologies de Lille, Villeneuve d’Ascq, France.
3
Biospheric Sciences Branch, NASA Goddard Space Flight Center,
Greenbelt, Maryland, USA.
Copyright 2006 by the American Geophysical Union.
0148-0227/06/2005JD006203$09.00
D05204 1of10

1999. A significantly low value of aerosol single-scattering
albedo at 550 nm (0.81 ± 0.08) was reported.
[
4] Given large advances in ground-based observation
techniques and analysis met hods, ground-based remote
sensing of aerosols is best suited to derive reliably and
persistently detailed aerosol properties in key locations of
the world, for example, urban aerosols in the mid-Atlantic
[Remer and Kaufman, 1998], biomass burning aerosols in
South Africa [Eck et al., 2003], dust aerosols in the Sahel
[Pinker et al., 2001], and maritime aerosols over the Pacific
and Atlantic oceans [Smirnov et al., 2002a]. These
researches have benefited from the rapid growth of AERO-
NET, a global ground-based network with the automatic
CIMEL Sun/sky radiometer as the principal instrument.
AERONET was initiated by the United States and France
in the early 1990s and expanded rapidly to over 100 stations
[Holben et al., 2001].
[
5] In March 2001, a CIMEL Sun/sky radiometer was
installed temporarily in Beijing as a part of the Aerosol
Characterization Experiment-Asia (ACE-Asia). Beijing be-
came one of the permanent AERONET/Photome´trie pour le
Traitement Ope´ rationnel de Normalisation Satellitaire
(PHOTONS) sites in April 2002. This paper will focus on
this heavily polluted and poorly understood region on the
basis of AERONET data. It is the first time that detailed
aerosol properties based on the 33-month AERONET data
in Beijing have been presented, as far as we know.
[
6] The paper is organized as follows. The experime ntal
site, measurement, and methodology are described in sec-
tion 2. Section 3 presents detailed results. Seasonal and
diurnal variations of aerosol properties are presented in
sections 3.1 and 3.2; sections 3.3 and 3.4 describe aerosol
size distribution and single-scattering albedo. The conclu-
sion and discussions are presented in section 4.
2. Site, Measurement, and Methodology
[7] Beijing is surrounded by mountainous area in the
west, the north, and the northeast. In the east and south of
Beijing, the elevation is close to sea level. There are four
distinct seasons. The cold and windy weather mainly occurs
in winter (December, January, and February) and spring
(March to May) because of frequent outbreaks of cold air
from west Siberia. Summer (June to August) in Beijing is
characterized by relatively hot and humid weather and
accounts for about 74% of annual precipitation. Autumn
(September to November) is generally a good season with a
relatively clear and clean sky. Winter is the heating season
that regularly begins in mid-November and ends in mid-
March. With a combination of gusting winds and loose
surface soil in the upwind Gobi and sandy regions, the dust
episode occasionally impacts Beijing in spring. There is not
a dust storm every day, but one or two at the very least can
be guaranteed in spring. Temperature inversion occurs
frequently in Beijing due partly to its special terrain,
especially in autumn and winter.
[
8] The radiometer was installed on the roof (above 30 m
height) of the Institute of Atmospheric Physics (IAP) office
building. In March 2004, the radiometer was moved to the
top of a building that is about 2 km away from IAP. The
automatic tracking Sun and sky scanning radiometer takes
measurements of solar direct radian ces and diffuse sky
radiances with a 1.2 full field of view. The direct Sun
measurements are made at five wavelengths (440, 670, 870,
940, and 1020 nm), and sky measurements are made at all
of the wavelengths except 940 nm. One set of measure-
ments requires about 10 s, and measurements are taken in
triplets at 30 s intervals. The direct Sun intensity measure-
ments are used to compute AOT at each wavelength except
for the channel at 940 nm, which is used for the retrieval of
column-integrated water vapor content. Angstrom exponent
is derived by AOT at 0.44, 0.67, and 0.87 nm. Aerosol size
distribution, refractive index, and single-scattering albedo
are retrieved from the sky radiance measurements and
AOTs. This is achieved by fitting the entire measured field
of radiances to a radiative transfer model that is driven by
the unresolved aer osol prop erties [Dubovik and King,
2000]. The AERONET data we present here are a level
2.0 quality-assured data set (data are available from the
AERONET Web site, http://aeronet.gsfc.nasa.gov/). The
data have been prefield and postfield calibrated, automati-
cally cloud screened [Smirnov et al., 2000], and manually
inspected. The uncertainty in AOT is 0.010.02 [Eck et al.,
1999]. Water vapor can be retrieved within 10% deviation.
The uncertainty in retrieval of radiative properties (refractive
index and single-scattering albedo) tends to increase as AOT
decreases. Column-integrated concentration (mm
3
/mm
2
) for
the particles of radius i n the range 0.1  7 mmcanbe
retrieved without significant error [Dubovik et al., 2000].
A spheroid particle shape assumption other than a spherical
assumption is used to simulate aerosol scattering during the
dust period, which is verified to be able to minimize
nonspherical effects of dust on retrievals, such as artificial
spectral dependence of the real part of the refractive index
and artificially high concentrations of small particles (radius
<0.1 mm) [Dubovik et al., 2002a].
[
9] Daily airborne particulate matter smaller than 10 mm
diameter (PM
10
) in Beijing is regularly measured by Ta-
pered Ele ment Oscillating Microbalance (TEOM)–based
analyzers in seven urban sites. The uncertainty of daily
PM
10
is estimated to be less than 1%. Air pollution index
(API) is then derived from the sum of daily PM
10
concen-
tration in seven sites using the following equation:
I ¼ I
high
 I
low

= C
high
 C
low

C  C
low
ðÞþI
low
; ð1Þ
where C is the PM
10
concentration (mg/m
3
). I
low
and
I
high
represent an API grading limited value that is lower
and larger, respectively, than the specified index I (API
index). C
high
and C
low
denote the PM
10
concentration
threshold for I
high
and I
low
(http://www.sepa.gov.cn/quality/
background.php).
[
10] API data are released online by the State Environ-
mental Protection A dministration of China (http://www.
sepa.gov.cn/quality/air.php3). On the basis of API and its
derivation equation (1), we compute daily PM
10
concentra-
tion using equation (2) if PM
10
is the principal pollutant.
C ¼ I  I
low
ðÞ
= I
high
 I
low

C
high
 C
low

þ C
low
: ð2Þ
3. Results
3.1. Seasonal Variation of Aerosol Properties
[
11] Figure 1a presents the monthly average of AOT at
440 nm, Figure 1b presents the Angstrom exponent,
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
2of10
D05204

Figure 1c presents surface PM
10
concentration, and
Figure 1d presents water vapor content. Monthly means
are derived from daily values. The daily values are com-
puted when daily observations exceed 10 (in order to make
the daily average representative given that diurnal variation
is large, as shown in section 3.2). Monthly AOT is higher in
spring and summer but relatively lower in autumn and
winter. The fact that the Angstrom exponent is the lowest
in spring indicates the huge impact of dust cases on aerosol
size in Beijing. Distinct monthly variation of water vapor is
evident. The largest water vapor occurs in summer, the
rainy season. Monthly water vapor contents in winter and in
spring are about 0.5 cm or less. Note that seasonal variation
of PM
10
is quite different from that of AOT, although both
of them ar e parameters cha racterizing aerosol loading.
Wi nter generally experiences high PM
10
concentration.
PM
10
remains high until April and then decreases steadily
to July. PM
10
increases gradually afterward. Two distinct
but quite different seasonal variations in AOT and PM
10
are
more evident as shown by Figure 2, in which a 4-year
monthly average of AOT and PM
10
is presented. Figure 3
presents scatterplots of daily AOT and PM
10
concentrations
in four seasons. AOT and PM
10
correlate significantly in
summer, indicating that aerosols are well mixed, and
surface measurement is a good indicator of column-inte-
grated value. In autumn and winter, AOT and PM
10
are also
well correlated. The correlation coefficient is 0.70 and 0.61,
respectively. A wide spread is evident in spring. The fact
that AOT and PM
10
correlated closely in Beijing is extraor-
dinary given that daily AOT and PM
10
are obtained under
different conditions. Daily PM
10
concentrations include
measurements under all-sky conditions and throughout the
whole day, but daily AOT is available only in the daytime
and under clear-sky conditions. Moreover, AOT is made up
of mostly fine particles, so PM
10
is not the optimal indicator
of AOT. Chu et al. [2003] also found a good correlation
between PM
10
and AOT in northern Italy (R = 0.82), but
this does not hold in Kanpur, a city in India. Singh et al.
[2004] reported a poor correlation (R = 0.24). AOT was
suggested to be a good indicator of PM
10
only in those
regions where pollution sources remained relatively invari-
able and air masses were stable. However, this poor
correlation is probably the result of the quite different
seasonal variation of AOT and PM
10
. From the statistical
view of point, AOT in summer is about two, three, and four
Figure 1. (a) Monthly mean of aerosol optical thickness at 440 nm, (b) Angstrom exponent from
AERONET, (c) surface PM
10
concentration (mg/m
3
) derived from Chinese State Environment Protection
Administration (SEPA) air pollution index (API) when PM
10
is the principal pollutant, and (d) monthly
mean water vapor from AERONET.
Figure 2. Bar graphs of 4-year monthly average of aerosol
optical thickness (AOT) at 440 nm and PM
10
concentration
(mg/m
3
). The quite different seasonal variation in AOT and
PM
10
concentration is evident.
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
3of10
D05204

times that in autumn, winter, and spring, respectively, for
the same PM
10
concentration. This indicates that more
aerosols are aloft in summer as compared with other
seasons. Fueled by strong solar radiation in summer, the
frequency of atmosphere inversion decreases, and its inten-
sity also becomes weak (see Table 1, in which monthly
frequency and intensity of temperature inversion are pre-
sented). Atmospheric convection is intensive as being the
heating of the atmosphere from below. Therefore aerosols
can be transported to a high level of the atmosphere. This
indicates the increase of atmosphere capability. The trans-
portation of aerosols to the troposphere can result in the
decrease of PM
10
. On the other hand, the column-integrated
aerosol loading will increase because more aerosols are
emitted into the column and maintain their existence with
relatively longer lifetimes. However, the temperature inver-
sion in spring and winter acts like a lid and prevents
pollution from penetrating it. As a result, aerosols accumu-
late in the surface, leading to a high concentration of PM
10
.
In addition, the effects of seasonal variation of aerosol
sources and meteorological factors should also play a role
in the observed different seasonal changes in AOT and
PM
10
. Figure 4 presents the streamline analysis in January
and July. It is shown that the dominant airflow to Beijing in
winter is from northwestern regions where AOT is close to
the background level. However, airflow to Beijing in
summer is from southern regions where AOT is greater
than 0.6. The advection of a high AOT pollution plume to
Beijing in summer results in more aerosols aloft in summer
than in winter. The season al variation of aerosol vertical
profile is verified by Lidar measure ments in Beijing. Hu et
al. [2004] found that the height of the pollutant boundary
layer in Beijing (the thickness between the surface and a
specified layer less than 2.5 km, where PM
10
is 50 mg/m
3
,
the value used to define air quality as the first class) was
about 1.36 km in summer; however the height was only
about 0.51 km in winter. Given the fact that mean PM
10
concentration in winter is 0.18 mg/m
3
and the value in
summer is 0.13 mg/m
3
, AOT in summer should be 1.86
times that in winter if we take the seasonal difference in
pollutant boundary layer into consideration. The estimated
AOT ratio of summer to winter is 1.90, which is close to the
observed value. This agreement indicates that the seasonal
difference in boundary layer should be one of the potential
Figure 3. Scatterplots of daily AOT at 440 nm and PM
10
concentration in four seasons. An identical
scale in the x and y axes is used in order to show the seasonal variation of relation between both
parameters.
Table 1. Monthly Frequency, Bottom Height, Thickness, and Intensity of Temperature Inversion in Beijing
a
Jan Feb Mar April May June July Aug Sept Oct Nov Dec
Frequency, % 68 82 77 40 65 31 45 52 77 55 63 48
Bottom height, m 57 54 79 121 156 222 221 184 76 76 55 100
Thickness, m 193 228 194 175 171 89 104 109 233 176 234 196
Intensity, C/100 m 1.68 1.33 1.45 0.83 0.83 0.67 0.58 0.47 0.98 1.13 1.08 1.01
a
Results are derived from daily radiosonde data in 2002.
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
4of10
D05204

reasons for the quite different seasonal variation of AOT
and PM
10
.
[
12] Heavy aerosol loading in Beijing is not only shown
by a few appearances of background level of aerosol
concentration (AOT at 500 nm less than 0.10) but also is
shown by frequent occurrences of high AOT in four
seasons. For example, more than 10% of daily AOT
exceeds 2.0 in summer (see Figure 5). Figure 6 presents
the scatterplots of daily AOT and the Angstrom exponent in
four seasons. Note that high AOT is always coupled with an
Angstrom exponent of approxi mately 1.2 except in spring.
The impact of spring dust weather on aerosol size is clearly
demonstrated by the fact that some AOTs are coupled with
Angstrom expon ents less than 0.6 in t his seaso n. It is
difficult to exclude cloud contamination on AOT observa-
tions since AOT values are close to cloud optical thickness
occasionally. The fact that large AOTs are always associated
with an Angstrom exponent representative of aerosol indi-
cates that cloud effect on AOTs in Beijing is negligible.
Moreover, AOTs and Angstrom exponents are very stable
throughout the day when AOT is high. Daily coefficient of
variation (COV), the ratio of standard variation to the daily
average, of AOT is about 10–20% when AOT is larger than
1.0. COV is only half of that when AOT is moderate (0.4–
0.5). With respect to the Angstrom exponent, COV remains
10% and does not vary with AOT.
3.2. Diurnal Variation of Aerosol Properties
[
13] Diurnal variati on of AOT is an important issue for
satellite aerosol validation and aerosol radiative forcing
calculation [Smirnov et al., 2002b; Kaufman et al., 2000].
At most urban/industrial AERONET sites, a prevailing
pattern of the AOT was shown to increase by 10–40%
during the day time [Smirnov et al., 2002b]. Figure 7
presents diurnal variation of AOT, Angst rom exponent,
and water vapor content in four seasons in Beijing. The
diurnal variability of AOT in Beijing is significant and
seasonally consistent. It varies from about 15% in summer
to about 45% in autumn. AOT steadily increases throughout
the daytime and reaches maximum in the late afternoon. It
should be noted that surface aerosol concentration in Bei-
jing generally exhibits peak value in the morning, then
steadily decreases and reaches minimum in the late after-
noon (about 1600 local standard time) [Wang et al., 2004;
He et al., 2001]. Aerosol scattering coefficient at the surface
exhibits a similar diurnal pattern to aerosol concentration
[Bergin et al., 2001]. This diurnal variation of aerosol
concentration at the surface can be partly attributed to the
diurnal variation of aerosol sour ces, for example, peak
values in the morning (rush time to work). B ut more
important, it should be related to the diurnal variation of
boundary layer and associated dilution of aerosols in the
troposphere. The boundary layer generally increases after
sunrise as a result of increasing temperature at the surface.
Therefore aerosol concentration at the surface would
decrease gradually along with aerosol transportation from
surface to troposphere. On the contrary, the column-
integrated aerosol concentration would increase because
more aerosols are transported into the troposphere along
with this process. Except in winter, the diurnal variation of
the Angstrom exponent (within 10%) is about a factor of 3
less than that of the AOT. Angstrom exponents in the early
morning and in the later afternoon are less than the daily
average. However, they exceed the daily average by a few
percent in midday. This indicates that aerosol particle size
increases slightly from morning until noon, then reverses to
decrease in the afternoon. This daily pattern resembles that
of solar insolation. Solar insolation is one of the key factors
controlling the production of fine secondary particles. More
new particles would be expected in midday because of
strong solar insolation, then resulting in a relatively larger
Angstrom exponent. The diurnal pattern of water vapor is
similar to that of AOT, but the diurnal variation of water
Figure 4. Moderate Resolution Imaging Spectroradi-
ometer (MODIS) –derived aerosol optical thickness at
550 nm and National Centers for Environmental Prediction
850 hPa streamline analysis for (a) January and (b) July.
The diamond in each plot is the location of Beijing.
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
5of10
D05204

vapor is less than 15%, much less than that of AOT.
Specifically, the diurnal variation of water vapor in summer
is negligible , which corresponds to t he lowest diurnal
variation of AOT.
3.3. Aerosol Size Distribution
[
14] Column-integrated aerosol volume size distribution
in Beijing can be fit well by a two-mode lognormal
distribution (see Figure 8). The remarkable feature is that
both volumes of two modes increase with AOT by a
magnitude of about 0.1 mm
3
/mm
2
per unit AOT except in
spring. The volume concentration of coarse mode in spring
increases with AOT by more than two times that derived in
the other three seasons. Table 2 presents the linear fit results
of regressing aerosol size parameters to AOT. Note that all
retrievals using a spheroid particle shape assumption other
Figure 5. Frequency distribution (percent) of daily AOT at 0.44 mm in four seasons. Note that the same
scale in the x and y axes is adopted in order to check the seasonal difference.
Figure 6. Scatterplot of daily AOT at 440 nm and Angstrom exponent in four seasons. The Angstrom
exponent is derived by linear regression of AOT at 440, 670, and 870 nm.
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
6of10
D05204

than a spherical assumption is deliberately labeled as dust
class. (Daily meteorological observations and positive Total
Ozone Mapping Spectrometer (TOMS) aerosol indexes
corresponding to these retrievals support this classification.)
The outstanding feature for the dust case is that volume
concentration of coarse mode increases dramatically with
AOT, that is, 0.71 mm
3
/mm
2
per unit AOT, whereas the
volume concentration of fine mode during the dust period
increases with AOT by the same magnitude as that derived
in other cases. Coarse mode radius and mode standard
deviations remain relatively stable in Beijing. As for fine
mode radius, it increases with AOT in four seasons, which
resembles that in other urban regions [Remer et al., 1998;
Dubovik et al., 2002b]. The difference is that the increased
magnitude in Beijing is only 0.030.04 per unit AOT. The
value is about one third that derived in Maryland (0.11) and
in the suburban region of Paris (0.13). The increased
magnitude in Beijing is similar to that derived in urban
regions of Mexico and India [Dubovik et al., 2002b; X. A.
Xia et al., unpublished manuscript, Evaluation of aerosol
optical thickness of MODIS onboard Terra and Aqua in
Asia, 2005]. Note that Maryland and Paris are humid
environments versus relatively dry environments in Beijing
and Mexico City, so aerosol swelling from increase of
humidity and aerosol aging in Maryland and Paris contrib-
utes to part of the AOT increase and aerosol size simulta-
neously. On the contrary, large input of new or fresh small
particles to the atmosphere in Beijing with relatively fixed
size from local sources or from outside is persistent, which
would be expected to account for a large fraction of the
AOT increase. Hence one would not expect aerosol size
increases with aerosol loading in Beijing and Mexico City
by the same magnitude as that in Maryland and Paris. It is
the submicron aerosols that are generally responsible for the
majority of the light scattering [Bergin et al., 2001; Xia et
al., 2005]. Since submicron aerosol sizes in urban regions
differ distinctly from each other, one would expect a
considerable difference in aerosol phase function, especially
within the side and backscattering angles that are regularly
used in satellite remote sensing of aerosol. This indicates
that different aerosol models should be established and used
in satellite remote sensing even for the same aerosol type.
The analysis of aerosol absorption in section 3.4 will also
support this suggestion.
3.4. Single-Scattering Albedo and Refractive Index
[
15] Much concern has been devoted to aerosol absorp-
tion in recent years [Hansen et al., 1997; Ramanathan et al.,
2001]. For nonabsorptive aerosols such as sulfate, the direct
aerosol forcing at the surface and top of the atmosphere
(TOA) are nearly identical. But when absorptive aerosols
such as black carbon (BC) are present, the reduction of the
surface insolation is the sum of the solar radiation reflected
back to space and absorbed in the atmosphere. In this case,
one would expect that the direct aerosol radiative forcing at
the surface is much larger than the forcing at the TOA.
Satheesh and Ramanathan [2000] demonstrated that the
direct radiative forcing over the Indian Ocean is about three
times higher than that at the TOA.
[
16] Ground sampling researches show that the mass
fraction of BC in Beijing varies from about 5% to 10%
[He et al., 2001; Sun et al., 2004]. The BC ratio in Beijing is
close to the result derived in the Indian Ocean Experiment
(INDOEX) [Ramanathan et al., 2001]. As a result, high
aerosol absorption would also be expected in Beijing.
Figure 9 presents averaged single-scattering albedo in four
seasons. Dust cases are deliberately grouped in an excep-
tional class and are also presented in Figure 9. Mean single-
scattering albedos at 440 nm in Beijing are about 0.90;
hence one would expect a similar or even larger discrepancy
in aerosol direct forcing at TOA and the surface given that
aerosols suspend over the bright land here. Single-scattering
albedo in Beijing is significantly less than that derived in
other urban regions; for example, single-scattering albedo at
440 nm is about 0.96 in Maryland and 0.93 in Paris.
Figure 7. Diurnal variability of aerosol optical thickness,
Angstrom exponent, and water vapor computed hourly as
percent departure from daily average.
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
7of10
D05204

However, it is close to that obtained in Mexico City (about
0.90 at 440 nm) and in Kanpur (about 0.90 at 440 nm)
[Dubovik et al., 2002b; Singh et al., 2004]. A remarkable
feature deserving mention is the fact that single-scattering
albedo increases moderately during the dust period, and the
spectral dependence also changes from decrease with wave-
length to the reverse. The real part of the refractive index
increases from 1.44 to 1.52 with wavelength from 440 to
1020 nm. The real part is slightly higher in winter and in
spring. Note that the real part increases during the dust
episode, and the spectral dependence of the real part also
shifts to the reverse because of the input of dust aerosols.
The imaginary part of the refractive index in winter reaches
about 0.014, decreases to about 0.012 in autumn, and is
about 0.009 in summer and spring. The decrease in imag-
inary index in summer is probably related to the hygro-
scopic growth in summer; as for the decrease in spring, it
resulted from dust input. The difference in refractive index
is remarkable between different urban regions; for instance,
the imaginary part in Maryland is about 0.003. The
difference in aerosol radiative properties is closely related
to different aerosol chemical and physical properties that
are determined by energy structure and industry type. In
addition, it should be noted that climate is also an
important factor affecting aerosol properties; for example,
Maryland and Paris are humid environments versus rela-
tively dry environments at Beijing and Mexico City, so
strong humidification of aerosol in Maryland and Paris
will result in less absorption. On the contrary, aerosol
loading is largely made up of a large input of new or fresh
small particles, with a large fraction of BC in Mexico City
and Beijing, so strong absorption would be expected.
4. Conclusions
[17] In this paper, we presented aerosol column-integrated
properties (loading, size distribution, and absorption) in
Beijing, a very polluted and highly populated megacity.
The analysis was based on 33 months of AERONET data. A
few conclusions were reached.
[
18] 1. Much higher aerosol loading than the background
level has been observed in Beijing. Monthly mean AOT at
440 nm in Beijing ranges from about 0.41 in January to
more than 1.25 in June. Seasonal variat ion of AOT is
distinct, with high monthly AOT in spring and summer
and relatively low monthly AOT in autumn and winter. This
Figure 8. Aerosol volume size distributions for AOT at 0.44 mm varying from about 0.10 to 1.60 in four
seasons. Each size distribution is an average compu ted from more than 10 instantaneous almucantar
retrievals.
Table 2. Results Derived From the Linear Regression of Aerosol Size Parameters to Aerosol Optical Thickness at 440 nm in Four
Seasons and During Exceptional Dust Period
a
Parameter Points V
f
,
b
mm
3
/mm
2
r
vf
,
b
mm s
vf
V
c
r
vc
s
vc
Spring 408 0.09x + 0.03 0.03x + 0.12 0.50 0.25x + 0.02 2.82 0.73
Summer 289 0.10x + 0.03 0.03x + 0.13 0.50 0.12x + 0.03 3.08 0.77
Autumn 316 0.09x + 0.04 0.04x + 0.11 0.49 0.12x + 0.04 2.96 0.71
Winter 304 0.10x + 0.03 0.03x + 0.12 0.54 0.14x + 0.02 3.05 0.75
Dust period 71 0.09x  0.01 0.13 0.73 0.72x  0.03 2.39 0.56
a
V
f
is mean volume concentration, r
vf
is volume radius, and s
vf
is volume deviation for find mode. The corresponding values for coarse mode are
represented by V
c
, r
vc
, and s
vc
.
b
The x refers to aerosol optical thickness at 440 nm.
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
8of10
D05204

seasonal pattern of AOT is quite different from that of the
surface PM
10
concentration. PM
10
decreases steadily from
its peak value in winter and early spring to its trough in
summer. This inconsistency is partly ascribed to the sea-
sonal variation of pollution boundary la yer height that is
determined by solar insolation and aerosol transportation.
Additionally, other factors such as seasonal variation of
relative humidity (aerosol hygroscopic growth) cannot be
excluded.
[
19] 2. The analysis shows a seasonally consistent
diurnal variation of AOT in Beijing. AOT increases nearly
monotonically during the daytime, and the diurnal vari-
ability varies from 15% in summer to 45% in winter. This
is quite different from diurnal variation of surface PM
10
concentration. The inconsistency is related to diurnal
variation of the pollution boundary layer height and
associated diurnal variation of aerosol dilution from the
surface to a high level.
[
20] 3. The relationship of PM
10
concentration to AOT in
summer is statistically significant, indicating that aerosols
are well mixed, and surface observation is a good indictor of
the column value. A good correlation between AOT and
PM
10
is de rived in autumn and winter. The correlation
coefficient is 0.70 and 0.61, respectively. A wide spread is
evident in spring. The different ratios of AOT to PM
10
have
been derived for different seasons as a result of the seasonal
variation of the height of pollutant layer.
[
21] 4. Aerosol volume size distribution in Beijing shows
two distinct modes. Volume concentrations of two modes
increase with AOT by a nearly identical magnitude except
in spring. Volume concentration of coarse mode in spring
increases with AOT by a bout two times that in other
seasons. Coarse mode radius and mode standard deviation
of two modes remains relatively stable with AOT. However,
fine mode radius in Beijing increases with AOT. Its increase
trend in Beijing is about one third that in Maryland and
Paris.
[
22] 5. Aerosol single-scattering albedo in Beijing is
about 0.90 at 440 nm and decreases slightly to the near-
infrared wavelength. Aerosol single-scattering albedo will
increase notably, and the spectral dependence will reverse if
Beijing is impacted by dust weather. Aerosol size and
absorption in Beijing is close to AERONET retrievals in
Mexico City and Kanpur, but they are remarkably different
from those retrieved in Maryland and Paris. This indicates
that different urban aerosol models should be created and
used in the spaceborne remote sensing of aerosols in
different urban regions.
[
23] Acknowledgments. The authors thank Chinese SEPA for pro-
viding API. This research is partly supported by the NSFC foundation
(40305002, 40333029, and 40520120071) and the China-Fra nce PRA
project. Special thanks go out to two anonymous reviewers for their
comments and suggestions.
References
Bergin, M. H., et al. (2001), Aerosol radiative, physical, and chemical
properties in Beijing during June 1999, J. Geophys. Res., 106,
17,969 –17,980.
Charlson, R. J., S. E. Schwartz, J. M. Hales, D. Cess, J. A. Coakley, and
J. E. Hansen (1992), Climate for cing by anthropogenic aeros ols,
Science, 255, 423 – 430.
Chu, D. A., Y. J. Kaufman, G. Zibordi, J. D. Chern, J. Mao, C. Li, and B. N.
Holb en (2003), Gl obal monitoring of air pollution over land from
EOS-Terra MODIS, J. Geophys. Res., 108(D21), 4661, doi:10.1029/
2002JD003179.
Dubovik, O., and M. D. King (2000), A flexible inversion algorithm for
retrieval of aerosol optical properties from Sun and sky radiance mea-
surements, J. Geophys. Res., 105, 20,673–20,696.
Dubovik, O., A. Smirnov, B. N. Holben, M. D. King, Y. J. Kaufman, T. F.
Eck, and I. Slutsker (2000), Accuracy assessments of aerosol optical
properties retrieved from Aerosol Robotic Network (AERONET) Sun
and sky radiance measurements, J. Geophys. Res., 105, 9791 –9806.
Dubovik, O., B. N. Holben, T. Lapyonok, A. Sinyuk, M. I. Mishchenko,
P. Yang, and I. Slutsker (2002a), Non-spherical aerosol retrieval method
employing light scattering by spheroids, Geophys. Res. Lett., 29(10),
1415, doi:10.1029/2001GL014506.
Dubovik, O., et al. (2002b), Variability of absorption and optical properties
of key aerosol types observed in worldwide locations, J. Atmos. Sci., 39,
590 –608.
Eck, T. F., B. N. Holben, J. S. Reid, O. Dubovik, A. Smirnov, N. T. O’Neill,
I. Slutsker, and S. Kinne (1999), Wavelength dependence of the optical
depth of biomass burning, urban and desert dust aerosols, J. Geophys.
Res., 104, 31,333 –31,350.
Eck, T. F., et al. (2003), Variability of biomass burning aerosol optical
characteristics in southern Africa during the SAFARI 2000 dry season
campaign and a comparison of single scattering albedo estimates
from radiometric measurements, J. Geophys. Res., 108(D13), 8477,
doi:10.1029/2002JD002321.
Eck, T. F., et al. (2005), Columnar aerosol optical properties at AERONET
sites in central eastern Asia and aerosol transport to the tropical mid-
Pacific, J. Geophys. Res., 110, D06202, doi:10.1029/2004JD005274.
Figure 9. Seasonal average of aerosol single-scattering
albedo and the real part and imaginary part of the refractive
index at four wavelengths (0.44, 0.67, 0.87, and 1.02 mm).
Also included are corresponding aerosol properties derived
during the dust period in Beijing.
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
9of10
D05204

Hansen, J., M. Sato, and R. Ruedy (1997), Radiative forcing and climate
response, J. Geophys. Res., 102, 6831 –6864.
He, K. B., F. M. Yang, Y. L. Ma, Q. Zhang, X. H. Yao, C. K. Chan,
S. Cadle, T. Chan, and P. Mulawa (2001), The characteristics of
PM
2.5
in Beijing, China, Atmos. Environ., 35, 4959 – 4970.
Holben, B. N., et al. (1998), AERONET—A federated instrument network
and data archive for aerosol characterization, Remote Sens. Environ., 66,
1 –16.
Holben, B. N., et al. (2001), An emerging ground-based aerosol clima-
tology: Aerosol optical depth from AERONET, J. Geophys. Res., 106,
12,067 –12,098.
Hu, H. L., Y. H. Wu, C. B. Xie, F. Q. Yan, N. Q. Weng, A. Y. Fan, C. D.
Xu, Y. F. Ji, T. Yu, and Z. H. Ren (2004), Aerosol pollutant boundary
layer measured by Lidar in Beijing, Res. Environ. Sci., 17, 59 – 73.
Husar, R. B., et al. (2001), Asian dust events of April 1998, J. Geophys.
Res., 106, 18,317 –18,330.
Kaiser, D. P., and Y. Qian (2002), Decreasing trends in sunshine duration
over China for 1954 –1998: Indication of increased haze pollution?,
Geophys. Res. Lett., 29(21), 2042, doi:10.1029/2002GL016057.
Kaufman, Y. J., B. N. Holben, D. Tanre´, I. Slutsker, A. Smirnov, and T. F.
Eck (2000), Will aerosol measurements from Terra and Aqua polar orbit-
ing satellites represent the daily aerosol abundance and properties?, Geo-
phys. Res. Lett., 27, 3861 –3864.
Kaufman, Y. J., D. Tanre´, and O. Bo ucher (2002), A satellite view of
aerosols in the climate system, Nature, 419, 215 – 223.
Li, X. W., X. L. Zhou, and W. L. Li (1995), The cooling of Sichuan
province in recent 40 years and its probable mechanism, Acta Meteorol.
Sin., 9, 257 –268.
Luo, Y. F., D. R. Lu, X. J. Zhou, W. L. Li, and Q. He (2001), Characteristics
of the spatial distribution and yearly variation of aerosol optical depth
over China in last 30 years, J. Geophys. Res., 106, 14,501 – 14,513.
Menon, S., J. E. Hansen, L. Nazarenko, and Y. F. Luo (2002), Climate
effects of black carbon aerosols in China and India, Science, 297, 2249 –
2252.
Pinker, R. T., G. Pandithurai, B. N. Holben, O. Dubovik, and T. O. Aro
(2001), A dust outbreak episode in sub-Sahel west Africa, J. Geophys.
Res., 106, 22,923 –22,930.
Qian, Y., and F. Giorgi (2000), Regional climatic effects of anthropogenic
aerosols? The case of sou thwester n China, Geophys . Res. Lett., 27,
3521 –3524.
Ramanathan, V., P. J. Crutzen, J. T. Kiehl, and D. Rosenfeld (2001), Aero-
sols, climate, and the hydrological cycle, Science, 294, 2119– 2124.
Remer, L. A., and Y. J. Kaufman (1998), Dynamic aerosol model: Urban/
industrial aerosol, J. Geophys. Res., 103, 13,859–13,871.
Satheesh, S. K., and V. Ramanathan (2000), Large differences in tropical
aerosol forcing at the top of the atmosphere and Earth’s surface, Nature,
405, 60 – 63.
Singh, R. P., S. Dey, S. N. Tripathi, V. Tare, and B. N. Holben (2004),
Variability of aerosol parameters over Kanpur, northern India, J. Geo-
phys. Res., 109, D23206, doi:10.1029/2004JD004966.
Smirnov, A., B. N. Holben, T. F. Eck, O. Dubovik, and I. Slutsker (2000),
Cloud screening and quality control algorithms for the AERONET data-
base, Remote Sens. Environ., 73, 337 – 349.
Smirnov, A., et al. (2002a), Optical properties of atmospheric aerosol in
maritime environments, J. Atmos. Sci., 59, 484 – 500.
Smirnov, A., B. N. Holben, T. F. Eck, I. Slutsker, B. Chatenet, and R. T.
Pinker (2002b), Diurnal variability of aerosol optical depth observed at
AERONET (Aerosol Robotic Network) sites, Geophy s. Res. Lett.,
29(23), 2115, doi:10.1029/2002GL016305.
Sun, Y. L., G. S. Zhuang, Y. Wang, L. H. Han, J. H. Guo, M. Dan, W. J.
Zhang, Z. F. Wang, and Z. P. Hao (2004), The air-borne particulate
pollution in Beijing—Concentration, composition, distribution and
sources, Atmos. Environ., 38, 5991–6004.
Wang, J. L., Z. Xie, Y. H. Zhang, M. Shao, L. M. Zeng, C. L. Cheng, X. F.
Xu, X. F. Zhao, and Y. J. Meng (2004), The research on the mass con-
centration characteristics of find particles in Beijing, Acta Meterol. Sin.,
62, 105 – 111.
Xia, X. A. (2004), Significant decline in solar radiation over China: Causes
and implications, Eos Trans. AGU, 85(17), Jt. Assem. Suppl., Abstract
A12A-04.
Xia, X. A., H. B. Chen, P. C. Wang, X. M. Zong, and P. Gouloub (2005),
Aerosol properties and their spatial and temporal variations over north
China in spring 2001, Tellus, Ser. B, 57, 28 – 39.
Xu, Q. (2001), Abrupt change of the midsummer climate in central east
China by the influence of atmospheric pollution, Atmos. Environ., 35,
5029 –5040.

B. Chatenet, and P. Goloub, LOA, Universite´ des S ciences et
Technologies de Lille, Cite´ scientifique, Bat P5, F-59655 Villeneuve
d’Ascq Cedex, France.
H. B. Chen, P. C. Wang, X. A. Xia, and W. X. Zhang, Institute of
Atmospheric Physics, Chinese Academy of Sciences, Beijing, 100029,
China. (xiaxiangao2000@yahoo.com)
T. F. Eck and B. N. Holben, Biospheric Sciences Branch, NASA Goddard
Space Flight Center, Greenbelt, MD 20771, USA.
D05204 XIA ET AL.: AEROSOL IN A CHINESE URBAN REGION
10 of 10
D05204