Variability of marine aerosol fine-mode
fraction and estimates of anthropogenic
aerosol component over cloud-free
oceans from the Moderate Resolution
Imaging Spectroradiometer (MODIS)
Article
Published Version
Yu, H., Chin, M., Remer, L. A., Kleidman, R. G., Bellouin, N.,
Bian, H. and Diehl, T. (2009) Variability of marine aerosol fine-
mode fraction and estimates of anthropogenic aerosol
component over cloud-free oceans from the Moderate
Resolution Imaging Spectroradiometer (MODIS). Journal of
Geophysical Research - Atmospheres, 114 (D10). ISSN 0148-
0227 doi: https://doi.org/10.1029/2008JD010648 Available at
https://centaur.reading.ac.uk/34635/
It is advisable to refer to the publisher’s version if you intend to cite from the
work. See Guidance on citing
.
Published version at: http://dx.doi.org/10.1029/2008JD010648
To link to this article DOI: http://dx.doi.org/10.1029/2008JD010648
Publisher: American Geophysical Union
All outputs in CentAUR are protected by Intellectual Property Rights law,
including copyright law. Copyright and IPR is retained by the creators or other
copyright holders. Terms and conditions for use of this material are defined in
Variability of marine aerosol fine-mode fraction and estimates of
anthropogenic aerosol component over cloud-free oceans from the
Moderate Resolution Imaging Spectroradiometer (MODIS)
Hongbin Yu,
1,2
Mian Chin,
2
Lorraine A. Remer,
2
Richard G. Kleidman,
2,3
Nicolas Bellouin,
4
Huisheng Bian,
1,2
and Thomas Diehl
1,2
Received 24 June 2008; revised 24 March 2009; accepted 30 March 2009; published 29 May 2009.
[1] In this study, we examine seasonal and geographical variability of marine aerosol
fine-mode fraction ( f
m
) and its impacts on deriving the anthropogenic component of
aerosol optical depth (t
a
) and direct radiative forcing from multispectral satellite
measurements. A proxy of f
m
, empirically derived from the Moderate Resolution Imaging
Spectroradiometer (MODIS) Collection 5 data, shows large seasonal and geographical
variations that are consistent with the Goddard Chemistry Aerosol Radiation Transport
(GOCART) and Global Modeling Initiative (GMI) model simulations. The so-derived
seasonally and spatially varying f
m
is then implemented into a method of estimating t
a
and
direct radiative forcing from the MODIS measurements. It is found that the use of a
constant value for f
m
as in previous studies would have overestimated t
a
by about 20%
over global ocean, with the overestimation up to 45% in some regions and seasons. The
7-year (2001–2007) global ocean average t
a
is 0.035, with yearly average ranging from
0.031 to 0.039. Future improvement in measurements is needed to better separate
anthropogenic aerosol from natural ones and to narrow down the wide range of aerosol
direct radiative forcing.
Citation: Yu, H., M. Chin, L. A. Remer, R. G. Kleidman, N. Bellouin, H. Bian, and T. Diehl (2009), Variability of marine aerosol
fine-mode fraction and estimates of anthropogenic aerosol component over cloud-free oceans from the Moderate Resolution Imaging
Spectroradiometer (MODIS), J. Geophys. Res., 114, D10206, doi:10.1029/2008JD010648.
1. Introduction
[2] With the implementation of multiwavelength, multi-
angle, and polarization measuring capabilities, current satel-
lite measurements can be used to categorize aerosol types in
terms of microphysical properties, such as particle size and
shape [e.g., Kahn et al., 2001; Tanre´etal., 2001; Higurashi
and Nakajima, 2002; Winker et al., 2007]. For example, the
fine-mode fraction, a measure of the contribution of fine-
mode aerosols to the aerosol optical depth (AOD or t), has
been obtained from enhanced satellite sensors (e.g., the
Moderate resolution Imaging Spectroradiometer (MODIS))
with improved data quality [Tanre´etal., 1997; Remer et al.,
2005]. Given that anthropogenic aerosols are predominately
fine-mode or in the submicron range, the fin e-mode fraction
in conjunction with the total aerosol optical depth can be used
as a tool for separating anthropogenic aerosol from dust
[Kaufman e t al.,2002].Kaufman et al. [2005a, 2005b]
further developed a quantitative method that uses MODIS
over-ocean retrievals in a consistent way to estimate the
anthropogenic component (e.g., originating from industrial
and urban pollution and biomass burning smoke) of aerosol
optical depth, t
a
, as follows:
t
a
¼ f f
d
ðÞt f
m
f
d
ðÞt
m
½= f
a
f
d
ðÞ; ð1Þ
where t and f respectively represents total aerosol optical
depth and fine-mode fraction retrieved directly from MODIS,
both at 550 nm. Subscripts a, d, and m denote anthropogenic,
dust, and marine aerosol components, respectively. Marine
aerosol optical depth t
m
is empirically determined to be a
constant of 0.06 [Kaufman et al., 2005a] or a function of
near-surface wind speed [Kaufman et al., 2005b]. The fine-
mode fractions for marine ( f
m
), anthropogenic ( f
a
), and dust
( f
d
) aerosol were assumed to be constant, which were then
derived from Terra MODIS Collection 4 measurements in
selected regions where the specific aerosol type predominates
and contributions of background aerosol are empirically
accounted for [Kaufman et al., 2005a]. Clearly this algorithm
does not assume that all fine-mode AOD comes from
anthropogenic contribution or anthropogenic AOD is
exclusively fine-mode. Contributions from natural aerosols
(dust and marine aerosol) to fine-mode AOD are empirically
accounted for. The essence of this algorithm is that the
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, D10206, doi:10.1029/2008JD010648, 2009
1
Goddard Earth Science and Technology Center, University of Mary-
land at Baltimore County, Baltimore, Maryland, USA.
2
Laboratory for Atmospheres, NASA Goddard Space Flight Center,
Greenbelt, Maryland, USA.
3
Science System and Applications, Inc., Lanham, Maryland, USA.
4
Met Office, Exeter, UK.
Copyright 2009 by the American Geophysical Union.
0148-0227/09/2008JD010648
D10206 1of11
MODIS data are used in a consistent way. The MODIS fine-
mode fractions could be different from ground-truth values as
a result of retrieval uncertainties [Anderson et al., 2005a;
Kleidman et al., 2005]. However, by using the fine-mode
fractions ( f, f
m
, f
a
, and f
d
) consistently from MODIS, one
should be able to separate the components of t
a
and t
d
better
than if using inconsistent values of fine-mode fraction from
other sources. Parallel to Kaufman et al. [2005a, 2005b],
Bellouin et al. [2005] developed a method that uses
measurement-based fine-mode fraction thresholds for
anthropogenic aerosol and sea salt aided by satellite-
observed absorbing aerosol index to separate anthropogenic
aerosol from dust and sea salt.
[
3] These satellite-based approaches have since inspired
the community to further explore the use of satellites to
quantify aerosol direct radiative forcing by anthropogenic
aerosol [e.g., Anderson et al., 2005b, Christopher et al.,
2006; Yu et al., 2006] and to estimate trans-boundary trans-
port of pollution aerosol [Rudich et al. , 2008; Yu et al., 2008].
This approach, along with improvements in data sets from
ground-based network and field campaigns [Bates et al.,
2006; Yu et al., 2006, and references therein] and dedicated
and coordinated efforts on aerosol modeling [Schulz et al.,
2006], has contributed to the reduced uncertainty of both the
aerosol direct radiative forcing and total anthropogenic
radiative forcing as assessed in the Intergovernmental Panel
on Climate Cha nge (IPCC) Fourth Assessment Report
[Haywood and Schulz, 2007].
[
4] Significant endeavors are needed to further investigate
this approach and explore the use of satellite measurements
for a better understanding of anthropogenic aerosol radiative
forcing. For example, the inherent assumptions in deriving
equation (1) need to be assessed and improved. In this study,
we examine the assumption of constant fine-mode fraction
for marine aerosol ( f
m
) and propose a self-consistent approach
to improve the characterization of seasonal and spatial
variations of f
m
. Over remote oceans, aerosols are generated
from bursting bubbles that inject sea salt particles, dimethyl-
sulfide (DMS) and organic matters into the marine boundary
layer. The DMS oxidation produces SO
2
and sulfates. The
organic particles and DMS-oxidized sulfates contribute to the
optical depth predominantly in the submicron range. The sea
salt aerosols have much broader size distributions, with mass
concentrated in the supermicron size range. The submicron
sea salt is, however, much more efficient in scattering the
solar radiation. As a result, the submicron sea salt constitutes
a significant contributor to the sea salt optical depth and also
an important component of fine-mode marine aerosol optical
depth [Bates et al., 2001]. The amount, composition, and size
of marine-generated aerosols should depend on a variety of
atmospheric a nd oceanic paramete rs, such as biological
activities of ocean, sea-surface temperature, ocean upwelling,
near-surface wind speed, atmospheric oxidizing capacity,
among others [O’Dowd et al., 2004; Leck and Bigg, 2005].
This complexity would result in large seasonal and geograph-
ical variations of f
m
, as can be inferred from some observa-
tions [e.g., Wilson and Forgan, 2002; Shinozuka et al., 2004].
[
5] In section 2, we derive f
m
from the Terra MODIS
Collection 5 (C5) data and discuss its seasonal and geo-
graphical variations in conjunction with the Goddard Chem-
istry Aerosol Radiation Transport (GOCART) and Global
Modeling Initiative (GMI) model sim ulations of marine
aerosol. The derived marine fine-mode fraction, which is
seasonally and geographically varying, is then utilized to
derive the anthropogenic aerosol optical depth from 2001 to
2007 MODIS observations. Section 3 examines the seasonal
and interannual variability of t
a
and its comparisons with
model simulation and previous studies. Major results and
conclusions are summarized in section 4.
2. Marine Aerosol Fine-Mode Fraction From
MODIS: Seasonal and Geographical Variations
[6] MODIS C5 aerosol retrievals with consistent algo-
rithms have recently become available [Remer et al., 2006,
2008; Levy et al., 2007]. Because values of the aerosol fine-
mode fraction are sensitive to details of the algorithm and
MODIS calibration, it warrants a reassessment of the fine-
mode fractions for anthropogenic, dust, and marine aerosol in
order to apply the method of Kaufman et al. [2005a, 2005b]
to MODIS C5. Jones and Christopher [2007], hereafter
referred to as JC07, derived the fine-mode fraction values
for anthropogenic, dust, and sea sal t aerosol from 1-year
Terra MODIS C5 data, with aerosol type characterization
guided by GOCART model simulations. Here we follow the
method as described by Kaufman et al. [2005a, 2005b] by
selecting the representative regions and seasons dominated
by pollution (i.e., Nort h Atlantic off the coast of New
England in summer), dust (i.e., North Atlantic off the coast
of North Africa in summer), and marine aerosol (i.e., south
to Australia) to the Terra MODIS C5 Level 3 daily data (at
aresolutionof1° 1°) from 2001 to 2007 . When
deriving fine-mode fractions for dust and pollution, a
contribution by marine aerosol is empirically excluded
[Kaufman et al., 2005a]. Table 1 lists the newly derived
representative values of fine-mode fraction for individual
aerosol types and their comparisons with those derived from
Terra MODIS Collection 4 (C4) data [Kaufman et al., 2005a,
2005b]. The fine-mode fraction for mineral dust as derived
from C5 is smaller than that from C4, while that for marine
aerosol shows the opposite relationship. For pollution aerosol,
the fine-mode fraction is similar between C5 and C4. These
differences result from assumptions about the optical prop-
erties of coarse-mode particles that were adjusted to better
match more recent observations in C5 algorithm [Remer et
al., 2008]. The consequence of this algorithm change is to
reduce the positive bias in the fine-mode fraction retrieved by
C4 [Kleidman et al., 2005; Remer et al., 2008].
[
7] Our derived fine-mode fractions values are different
from those derived by JC07. JC07 derived fine-mode fraction
is 0.83, 0.44, and 0.25 for anthropogenic, dust, and sea salt
aerosol, respectively. Differences between this study and that
of JC07 could have resulted from several possible factors.
First, JC07 derived these values over much broader areas than
this study does and the differences between the two studies
may have come from spatial variability of particle size for a
specific aerosol type. It is also possible t hat MODIS and
GOCART differ in the characterization of aerosol types.
Second, while JC07 used monthly MODIS C5 data, we are
using MODIS C5 daily 1° 1° data in this study. Third, it is
expected that the fine-mode fraction for sea salt of JC07 is
smaller than the derived value for marine aerosol in this study
D10206 YU ET AL.: ANTHROPOGENIC AEROSOL FROM SATELLITE
2of11
D10206
because of the exclusion of contribution of submicron sulfate
produced from DMS by JC07. Note also that the sea salt fine-
mode fraction from JC07 may be biased low because of cloud
contamination [Zhang et al., 2005] and unaccounted-for
whitecaps over the ‘‘roaring forties.’’ Fourth, contribution
of background marine aerosol was accounted for in this study
but not accounted for by JC07 when deriving the fine-mode
fractions for anthropogenic aerosol and dust. This difference
may partly explain larger fine-mode fraction for anthropo-
genic aerosol derived in this study.
[
8] By applying equation (1) with these new values of
f
m
, f
d
, and f
a
(in Table 1) to the MODIS C5 data, we
obtain the global ocean average anthropogenic AOD of
0.040, which is about 20% larger than the average (0.033 )
or is at the upper bound (with an estimated uncertainty of
30%) as derived from MODIS C4 [Kaufman et al., 2005a].
This is opposite to the study by Bellouin et al. [2008] that
show a 25% decrease of anthropogenic AOD over ocean
when updating from C4 to C5. Since the same thresholds
of the fine-mode fraction were used to separate aerosol
types for both C5 and C4 in Bellouin et al. [2005, 2008],
the reduced fine-mode fraction over oceans in C5 [Remer
et al., 2008] results in the smaller anthropogenic AOD
derived from C5 [Bellouin et al., 2008] than that from C4
[Bellouin et al., 2005].
[
9] As discussed earlier, mari ne aerosol fine-mode frac-
tion should present large spatial and seasonal variations and
a use of constant f
m
could introduce large errors to the
derived anthropogenic aerosol optical depth and direct radi-
ative forcing. Here we derive the climatology of seasonal
average fine-mode fraction for background marine aerosol by
averaging 2001 –2007 Terra/MODIS daily fine-mode frac-
tion weighted by t for 0.03 < t < 0.10 in individual 1° 1°
grids. We assume that f
m
has relatively small inter-annual
variability and the multiyear data are then used to obtain a
better spatial coverage. It is also required that the number of
available daily measurements in each 1° 1° grid box during
a season is no less than 10 for calculating a seasonal average.
The lower bound of t is set to exclude data with relatively
large uncertainties, while the upper bound of t is chosen
for a compromise of excluding continental influences but
acquiring adequate spatial coverage. Note that t < 0.03
accounts for 10% of all over-ocean data [Remer et al.,
2008]. Spatial gaps in the derived f
m
shrink or ex pand
respectively with increasing or decreasing the upper bound
of t near coasts of major continental aerosol source regions.
For the selected t range, continental influences are likely
to exist. However, such residual continental influence
appears to have a small effect on the derived f
m
. As shown
in Figure 1, for example, by varying the upper bound of t
from 0.1 to 0.15, the difference of the derived f
m
in individual
grids is predominantly (98.5%) within ±0.1, in which
80% is within ±0.05.
[
10] Kaufman et al. [2001] derived the climatology of
optical depth and properties of the baseline marine aerosol
from multiyear measurements of nine Aerosol Robotic
Network (AERONET) stations. The f
m
at 500 nm is about
0.64 over the Atlantic (averaged over five stations) and 0.56
over the Pacific (averaged over four stations) [Kaufman et
al., 2001, Table 2]. By extracting values from the derived
f
m
climatology in this study for those stations, we get f
m
at
550 nm of 0.58 ± 0.11 over the Atlantic and 0.54 ± 0.05
over the Pacific, which is in good agreement with that of
Kaufman et al. [2001]. Given the difficulty in obtaining
spatial and temporal variations of the fine-mode fraction of
marine aerosol from other measurements, we examine sim-
ulations of two globa l chemical transport models in this
study, namely GOCART and GMI. To model marine aerosol,
we only consider emissions of sea salt and DMS from ocean
and volcanic SO
2
. The sulfate produced from DMS and
volcanic emissions and sub-micron sea salt are categorized
into ‘‘fine mode’’ in calculating the fine-mode fraction for the
marine aerosol. GOCART takes into account both eruptive
and noneruptive volcanic sources [Chin et al., 2000a, 2002],
whereas GMI only includes noneruptive volcanoes. Pro-
cesses represented in the models are chemistry, convection,
advection, boundary layer mixing, dry and wet deposition,
gravitational settling, and hygroscopic growth of aerosol
particles. Despite being driven by same meteorological and
photochemical fields, the two models differ in the oceanic
DMS and volcanic SO
2
emissions, and parameterizations of
several processes that determine aerosol such as aqueous
phase reactions, dry and wet deposition, gravitational settling,
and convective transport [Chin et al., 2000a, 2000b, 2002;
Bian et al., 2009]. Thus the simulated marine fine-mode
fraction can still be different between the two models.
[
11] Figure 2 shows the MODIS-derived f
m
and its
comparisons with GOCART and GMI simulations. Differ-
ences of 0.1– 0.2 exist between the MODIS-based f
m
and
model simulations in some regions, which however would
generally fall within the uncertainty ranges of either method.
All three sets of f
m
show generally consistent, pronounced
seasonal and geographical variations. The marine fine-mode
fraction is larger in summer than in winter, and also larger
in tropical and coastal regions than in high latitudes and
remote oceans. These seasonal and spatial variations result
Figure 1. Frequency distribution of the f
m
difference
between using AOD at 550 nm of 0.15 and 0.10 as the
upper bound in deriving the marine aerosol fine-mode
fraction.
Table 1. Comparisons of the Fine-Mode Fractions at 550 nm for
Individual Aerosol Types Derived From Terra MODIS Collections 4
and 5
Aerosol Types Collection 5 Collection 4
Pollution 0.90 0.92
Mineral dust 0.37 0.51
Marine aerosol 0.45 0.32
D10206 YU ET AL.: ANTHROPOGENIC AEROSOL FROM SATELLITE
3of11
D10206
