Investigating size-segregated sources of elemental composition of particulate matter in the South China Sea during the 2011 Vasco cruise

Abstract

The South China Sea (SCS) is a receptor of numerous natural and anthropogenic aerosol species from throughout greater Asia. A combination of several developing countries, archipelagic and peninsular terrain, a strong Asian monsoon climate, and a host of multi-scale meteorological phenomena make the SCS one of the most complex aerosol–meteorological systems in the world. However, aside from the well-known biomass burning emissions from Indonesia and Borneo, the current understanding of aerosol sources is limited, especially in remote marine environments. In September 2011, a 2-week research cruise was conducted near Palawan, Philippines, to sample the remote SCS environment. Size-segregated aerosol data were collected using a Davis Rotating Uniform size-cut Monitor (DRUM) sampler and analyzed for concentrations of 28 elements measured via X-ray fluorescence (XRF). Positive matrix factorization (PMF) was performed separately on the coarse, fine, and ultrafine size ranges to determine possible sources and their contributions to the total elemental particulate matter mass. The PMF analysis resolved six sources across the three size ranges: biomass burning, oil combustion, soil dust, a crustal–marine mixed source, sea spray, and fly ash. Additionally, size distribution plots, time series plots, back trajectories and satellite data were used in interpreting factors. The multi-technique source apportionment revealed the presence of biogenic sources such as soil dust, sea spray, and a crustal–marine mixed source. Anthropogenic sources were also identified: biomass burning, oil combustion, and fly ash. Mass size distributions showed elevated aerosol concentrations towards the end of the sampling period, which coincided with a shift of air mass back trajectories to southern Kalimantan. Covariance between coarse-mode soil dust and fine-mode biomass burning aerosols were observed. Agreement between the PMF and the linear regression analyses indicates that the PMF solution is robust. While biomass burning is indeed a key source of aerosol, this study shows the presence of other important sources in the SCS. Identifying these sources is not only key for characterizing the chemical profile of the SCS but, by improving our picture of aerosol sources in the region, also a step forward in developing our understanding of aerosol–meteorology feedbacks in this complex environment.

Full text

Atmos. Chem. Phys., 20, 1255–1276, 2020
https://doi.org/10.5194/acp-20-1255-2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.
Investigating size-segregated sources of elemental composition
of particulate matter in the South China Sea during the
2011 Vasco cruise
Miguel Ricardo A. Hilario1,a, Melliza T. Cruz2,3, Maria Obiminda L. Cambaliza1,2, Jeffrey S. Reid4, Peng Xian4,
James B. Simpas1,2, Nofel D. Lagrosas2,b, Sherdon Niño Y. Uy2,c, Steve Cliff5, and Yongjing Zhao5
1Department of Physics, Ateneo de Manila University, Quezon City, Philippines
2Manila Observatory, Ateneo de Manila University campus, Quezon City, Philippines
3Institute of Environmental Science and Meteorology, University of the Philippines, Diliman, Quezon City, Philippines
4Marine Meteorology Division, Naval Research Laboratory, Monterey, CA, USA
5Air Quality Research Center, University of California, Davis, Davis, CA, USA
anow at: Manila Observatory, Ateneo de Manila University campus, Quezon City, Philippines
bnow at: Center for Environmental Remote Sensing, Chiba University, Chiba, Japan
cnow at: School of Engineering and the Built Environment, Birmingham City University, Birmingham, England
Correspondence: Maria Obiminda L. Cambaliza (mcambaliza@ateneo.edu)
Received: 12 April 2019 – Discussion started: 24 May 2019
Revised: 19 November 2019 – Accepted: 12 December 2019 – Published: 4 February 2020
Abstract. The South China Sea (SCS) is a receptor of
numerous natural and anthropogenic aerosol species from
throughout greater Asia. A combination of several develop-
ing countries, archipelagic and peninsular terrain, a strong
Asian monsoon climate, and a host of multi-scale meteo-
rological phenomena make the SCS one of the most com-
plex aerosol–meteorological systems in the world. However,
aside from the well-known biomass burning emissions from
Indonesia and Borneo, the current understanding of aerosol
sources is limited, especially in remote marine environments.
In September 2011, a 2-week research cruise was conducted
near Palawan, Philippines, to sample the remote SCS envi-
ronment. Size-segregated aerosol data were collected using
a Davis Rotating Uniform size-cut Monitor (DRUM) sam-
pler and analyzed for concentrations of 28 elements mea-
sured via X-ray fluorescence (XRF). Positive matrix factor-
ization (PMF) was performed separately on the coarse, fine,
and ultrafine size ranges to determine possible sources and
their contributions to the total elemental particulate matter
mass. The PMF analysis resolved six sources across the three
size ranges: biomass burning, oil combustion, soil dust, a
crustal–marine mixed source, sea spray, and fly ash. Addi-
tionally, size distribution plots, time series plots, back tra-
jectories and satellite data were used in interpreting factors.
The multi-technique source apportionment revealed the pres-
ence of biogenic sources such as soil dust, sea spray, and a
crustal–marine mixed source. Anthropogenic sources were
also identified: biomass burning, oil combustion, and fly ash.
Mass size distributions showed elevated aerosol concentra-
tions towards the end of the sampling period, which coin-
cided with a shift of air mass back trajectories to southern
Kalimantan. Covariance between coarse-mode soil dust and
fine-mode biomass burning aerosols were observed. Agree-
ment between the PMF and the linear regression analyses
indicates that the PMF solution is robust. While biomass
burning is indeed a key source of aerosol, this study shows
the presence of other important sources in the SCS. Iden-
tifying these sources is not only key for characterizing the
chemical profile of the SCS but, by improving our picture of
aerosol sources in the region, also a step forward in develop-
ing our understanding of aerosol–meteorology feedbacks in
this complex environment.
Published by Copernicus Publications on behalf of the European Geosciences Union.

1256 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
1 Introduction
In the midst of several developing countries, the South China
Sea (SCS) is a receptor for a multitude of natural and an-
thropogenic sources of aerosol. At the same time, the re-
gion exhibits some of the world’s most complicated mete-
orology due to its archipelagic and peninsular terrain and
strong Asian monsoon climate. Thus, the SCS hosts one of
the world’s most complex and sensitive composition and cli-
mate regimes (Balasubramanian et al., 2003; Yusuf and Fran-
cisco, 2009; Atwood et al., 2013a, b; Reid et al., 2012, 2013,
2015). It is known to be impacted not only by dust storms
and industrial pollution from China (Wang et al., 2011; At-
wood et al., 2013a) but also by biomass burning emissions
from the Maritime Continent (Balasubramanian et al., 2003;
Lin et al., 2007; Cohen et al., 2010a, b; Wang et al., 2011;
Reid et al., 2013, 2015, 2016). The transport of such emis-
sions is enabled by the long atmospheric residence times of
fine particles (Cohen et al., 2010a), potentially creating re-
gional and global concerns through their effects on radiative
forcing (Nakajima et al., 2007; Boucher et al., 2013; Lin et
al., 2014; Ge et al., 2014) and cloud properties (Sorooshian
et al., 2009; Lee et al., 2012; Boucher et al., 2013; Ross et
al., 2018).
Highlighting the unique combination of terrain and sea
that feeds into the complexity of the meteorological environ-
ment of the region, Reid et al. (2012) and Xian et al. (2013)
posed the long-range hypothesis that monsoonal flows and
higher-frequency meteorological phenomena are a major fac-
tor in seasonal aerosol dispersion. Biomass burning plumes
are known to cause severe haze episodes due to these mon-
soonal flows, raising concentrations of particulate matter
(PM) to impact cloud physics and, in some cases, to danger-
ous air quality levels across large areas, particularly in asso-
ciation with positive phases of the El Niño–Southern Oscil-
lation (ENSO) (Engling et al., 2014; Fujii et al., 2015). Like-
wise, biomass burning is a significant contributor to the re-
gion’s cloud condensation nuclei (CCN) budget in all years,
as are the region’s significant anthropogenic emissions (Bala-
subramanian et al., 2003; Field et al., 2008; Reid et al., 2012,
2013, 2015, 2016; Atwood et al., 2017).
Partly due to the emphasis on dramatic biomass burning as
the primary source of aerosol particles in the region, the con-
tributions of other regional sources are not well understood
or perhaps underappreciated. As the SCS is host to major
population centers, industry, major ports, and coal and oil
combustion are expected to be an important regional source
of aerosol particles in the Maritime Continent (MC). Coarse-
mode dust and biogenic particles may also play a role as ice
nuclei (O’Sullivan et al., 2014), as biomass burning plumes
are known to entrain such particles (Reid et al., 1994, 2005;
Schlosser et al., 2017). As such, a network of interacting
sources exists in the region surrounding the SCS, wherein
aerosol particles mix during transport and complicate source
apportionment. Understanding the nature of sources in the
remote MC and their contributions is key to characterizing
the aerosol environment in the SCS and its relationship with
cloud behavior and precipitation patterns in the region; this
is particularly true given the higher sensitivity of clouds to
particle perturbations at lower concentrations. However, the
source apportionment of aerosol particles is complicated by
their complex chemistry and interactions with the marine en-
vironment (Atwood et al., 2013a, 2017).
As part of the Seven SouthEast Asian Studies program (7-
SEAS), a research cruise (Reid et al., 2015) was conducted
in late September 2011 on board the Philippine-flagged M/Y
Vasco in the vicinity of the northern Palawan archipelago.
The goal of this cruise was to observe the behavior of aerosol
particles in the SCS and to test the transport hypothesis pro-
posed in Reid et al. (2012) that the Philippines is a long-range
receptor of aerosol species transported across the SCS during
the Asian summer monsoon from Borneo, Sumatra, and the
Malay Peninsula. In particular, the cruise aimed to observe
that emissions from the Maritime Continent were reaching
the southwest monsoon trough. The Palawan archipelago is
a good receptor site for regional emissions due to its largely
rural settlements and its location upwind relative to the rest of
the Philippines. The sampling period coincided with the pas-
sage of one tropical storm and two tropical cyclones (TCs).
Of particular importance is the passage of TC Nesat begin-
ning on 26 September 2011, as TC inflow arms are known to
cause abrupt changes in regional flows.
As part of the 2011 Vasco cruise, particulate matter was
collected using a size-segregated Davis Rotating Uniform
size-cut Monitor (DRUM) impactor analyzed for elemental
composition. While Reid et al. (2015) noted the presence of
smoke plumes in two episodes during the cruise, their ini-
tial analysis of the region’s atmospheric chemistry also sug-
gested the events were a mix of biomass burning and oil
or shipping emissions, due to elevated levels of vanadium.
Additionally, differences in elemental ratios, mass fractions,
and back trajectory origins between the two events support
the presence of other sources besides biomass burning. From
the initial analysis of aerosol chemistry presented by Reid et
al. (2015), this study aims to identify aerosol sources in the
SCS, highlight the source variability present in the region,
and further develop the current understanding of the effect
of regional meteorological phenomena on aerosol dispersion.
The paper shows that, although biomass burning is a major
source of aerosols in the SCS, anthropogenic sources such
as oil combustion also play an important role in the chem-
ical profile of the region. As we report, soil transport was
observed as well.
In this paper we expand on the original 2011 Vasco cruise
analysis to quantitatively apportion sampled biomass burning
and anthropogenic aerosol species. Positive matrix factoriza-
tion (PMF) was performed on size-segregated, elemental PM
to detect possible size-specific sources (Han et al., 2006; van
Pinxteren et al., 2016). Indeed, the relationship between the
aerodynamic diameter of a particle and its source has been
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1257
Figure 1. Path taken by the M/Y Vasco on 17–20 September (red),
20–28 September (black), and 28–30 September (blue). The ma-
jority of sampling was done at the northern end of Palawan island.
Image courtesy of Google Maps (map data ©2018 Google).
well-established in the literature (Reid et al., 1994; Balasub-
ramanian et al., 2003; Han et al., 2006; Lestari et al., 2009;
Wimolwattanapun et al., 2011; Santoso et al., 2011; Karana-
siou et al., 2009; Seneviratne et al., 2010; Atwood et al.,
2013a; Lin et al., 2015; Cahill et al., 2016). Aerosol factors
and characteristics were then used to spawn back trajectories
to identify individual island emissions areas.
2 Sampling and methods
2.1 Overall cruise sampling and environment
A general overview of the 2011 cruise can be found in Reid
et al. (2015), and a brief summary is provided here. Sampling
was conducted around the Palawan archipelago, an island
chain located at the southwestern edge of the Philippines in
between the SCS and the Sulu Sea. Sampling was performed
between Manila and the northern tip of Palawan Island on
board the M/Y Vasco, which left Manila Bay on 17 Septem-
ber 2011 and returned on 30 September 2011 (Fig. 1). The
majority of samples were collected around the areas of El
Nido and Malampaya Sound (111.1◦N, 119.3◦E), where the
vessel was stationed from 21 to 28 September. The largely
rural population of Palawan made it an ideal receptor for re-
gional rather than local emissions.
The cruise was conducted at the end of the Asian summer
monsoon, which usually lasts from June through September
(Loo et al., 2014; Chang et al., 2005). The Asian monsoon is
caused by the annual progress of the sun and asymmetrical
heating of air masses due to the complex terrain of Southeast
Asia (Chang et al., 2005). The campaign coincided with the
peak burning season in southern Kalimantan and southern
Sumatra, which have been measured to be the highest emit-
ters of biomass burning plumes in the MC (Reid et al., 2012).
As the southwest monsoon is characterized by winds travel-
ing southwest to northeast, Reid et al. (2015) proposed that
the Philippines was an excellent receptor for regional emis-
sions from the MC.
Although 2011 was a moderate La Niña year, it was noted
that fire activity and precipitation levels resembled a neu-
tral year (Reid et al., 2015). The cruise took place when the
Madden-Julian Oscillation (MJO) was transitioning from the
wet phase to the dry phase, which is expected to enhance
burning activity and transport. With the passage of tropical
cyclones (TCs), significant aerosol events were observed as
propagating across the region.
Reid et al. (2015) described three tropical events that oc-
curred during the cruise, specifically Tropical Storm (TS)
Haitang, TC Nesat, and Super Typhoon Nalgae. The pres-
ence of inflow arms in the SCS has been suggested to affect
the aerosol environment by bringing more MC air into the re-
gion (Reid et al., 2015). The passage of Nesat was observed
to abruptly affect air mass trajectories, coinciding with an
enhancement of several elements during the last 2 d of the
cruise.
Figure 2 shows the evolution of the meteorological envi-
ronment over the cruise period with comparisons between
aerosol optical depth (AOD) derived from the Moderate Res-
olution Imaging Spectroradiometer (MODIS) on board the
Terra and Aqua satellites, back trajectories from NOAA Hy-
brid Single Particle Lagrangian Integrated Trajectory Model
(HYSPLIT), and 850 hPa smoke concentrations from the
Navy Aerosol Analysis and Prediction System (NAAPS).
Back trajectories were run for 72 h, ending at 00:00 coordi-
nated universal time (UTC), i.e., 08:00 local time (LT), and
constrained isobarically, 300m above ground level (a.g.l.).
2.2 Aerosol sampling and analysis
Size-resolved aerosol samples were collected during the
cruise using a Davis Rotating Uniform size-cut Monitor
(DRUM) continuously sampling cascade impactor. Samples
were collected with a 10 µm inlet and eight size cuts at 5,
2.5, 1.15, 0.75, 0.56, 0.34, 0.26, and 0.10 µm at a 90 min
time resolution from midday 17 September until midday
30 September local time. Particles were collected on My-
lar strips coated with Apiezon grease. The eight drums were
rotated at a consistent rate to create a temporal record of
mass concentration (Raabe et al., 1988). X-ray fluorescence
(XRF) was performed on the DRUM samples at the Ad-
vanced Light Source (ALS) of Lawrence Berkeley National
Laboratory to measure mass concentrations of 28 elements,
ranging from Na to Pb. In this study, data were filtered based
on location notes from the cruise such that samples col-
lected in the vicinity of Manila Bay were excluded from the
analysis. Additionally, samples during an 8h pump failure
that occurred on 20 September were also excluded from the
dataset. In the analysis, the stages were aggregated into three
modes: coarse (1.15–10 µm), fine (0.34–1.15 µm), and ultra-
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1258 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
Figure 2. Satellite images of the SCS region taken from (a–d) NASA Worldview with an overlaid AOD; (e–h) NAAPS smoke concentration
plots (µg m−3; 850 hPa); and (i–l) HYSPLIT ensemble back trajectories during the cruise for 17, 22, 26, and 29 September (isobaric;
300 ma.g.l.; 72 h; ending at 00:00 UTC, i.e., 08:00LT). The red star indicates location of the Vasco.
fine (0.10–0.34 µm) modes. A large difference in the concen-
trations of stage 6 (0.34–0.56 µm) compared to (the adjacent)
stages 5 (0.56–0.75 µm) and 7 (0.26–0.34µm) was observed.
The sharp decrease in concentrations in stage 6 despite the
high concentrations in stages 5 and 7 has been observed in
other studies involving the DRUM sampler; this is likely due
to DRUM sampling artifacts and does not reflect the true
aerosol mass distribution (Atwood et al., 2013a). Neverthe-
less, the two size-resolved modes lend themselves to size-
segregated analysis. In this study, we simply report the mass
distributions as sampled by the DRUM.
In addition to the DRUM sampler, eight sets of PM2.5fil-
ters were collected during the cruise and were chemically an-
alyzed for information on species such as sulfate, nitrate, and
organic carbon. The PM2.5filters were described more fully
in Reid et al. (2015). Mass reconstruction was performed on
the PM2.5filter data according to the methodology of Malm
and Hand (2007). Results are shown in Fig. S1 and discussed
briefly in Sect. 3.1.
2.3 Model and satellite data
NOAA HYSPLIT model back trajectories (Draxler et al.,
1998, 1999; Rolph et al., 2017; Stein et al., 2015) were gen-
erated throughout the cruise period to investigate locations of
aerosol emission. HYSPLIT back trajectories have been used
in several studies to establish air mass source regions (Lin et
al., 2007; Cohen et al., 2010a; Atwood et al., 2013a, 2017).
Back trajectories were run for 72 h for heights of 500 and
300 m to investigate possible vertical inhomogeneity that has
been noted in other SCS papers (Atwood et al., 2013a). Tra-
jectory endpoints corresponded to cruise coordinates. Trajec-
tories were constrained isobarically to limit vertical wind ve-
locity since our area of interest is surface-level emission.
The NAAPS reanalysis product (Lynch et al., 2016), with
driving meteorology from the Navy Global Environmental
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1259
Model (NAVGEM), was used to provide overall aerosol and
meteorological context to the analysis. This reanalysis uti-
lizes a modified version of the NAAPS as its core and assim-
ilates quality-controlled retrievals of aerosol optical depth
(AOD) from MODIS on Terra and Aqua and the Multi-angle
Imaging SpectroRadiometer (MISR) on Terra (Zhang et al.,
2006; Hyer et al., 2011; Shi et al., 2014). NAAPS char-
acterizes anthropogenic and biogenic fine aerosols (includ-
ing sulfate, and primary and secondary organic aerosols), as
well as dust, biomass burning smoke, and sea salt aerosols.
Smoke from biomass burning is derived from near-real-
time satellite-based thermal anomaly data to construct smoke
source functions (Reid et al., 2009) with additional orbital
corrections on MODIS-based emissions and regional tuning.
The system has been successfully used to monitor biomass
burning plumes and to study the relationship of the aerosol
life cycle to weather systems over the MC (Reid et al., 2012,
2015, 2016; Atwood et al., 2013b; Xian et al., 2013).
Active fire hotspot data were downloaded from the
Fire Information for Resource Management System
(FIRMS) (https://firms.modaps.eosdis.nasa.gov/, last access:
19 July 2017). Active fire hotspots and aerosol optical depth
(AOD) at a wavelength of 550 nm were tracked throughout
the cruise via MODIS. MODIS detects thermal anomalies
across a region to identify possible fire activity. MODIS-
derived AOD was used to derive large-scale estimates of
PM2.5in some studies (e.g., Zheng et al., 2017). In the study,
MODIS was used to track burning emissions that were
found to be particularly prevalent in eastern Malaysia and
Indonesia. The use of MODIS to track active fire hotspots
has been used in other studies to understand seasonal trends
in agricultural burning (Reid et al., 2012) and to identify
and locate burning-related sources when used in conjunction
with HYSPLIT back trajectories (Atwood et al., 2017).
The NASA Worldview site (https://worldview.earthdata.
nasa.gov/, last accessed: 28 April 2018), an application oper-
ated by NASA’s Goddard Space Flight Center Earth Science
Data and Information System (ESDIS) project, was used to
supplement the satellite data by providing true color images
of the region and is particularly useful in demonstrating sud-
den changes of cloud environment or monsoon flow caused
by tropical cyclones.
2.4 Positive matrix factorization
Positive matrix factorization (PMF) was used to study the
covariance of elemental species. PMF is a multivariate fac-
tor analysis technique used in source apportionment that re-
solves a sample matrix X(i×j) of isamples and jspecies
into matrices G(i×k),F(k×j ), and E(i×j ), i.e., the source
contribution matrix, source profile matrix, and residual ma-
trix, respectively, with the assumption of kfactors as follows:
Xij =Gik Fkj +Eij .
The goal of PMF is to determine the number of factors or
sources ksuch that the solution will be physically inter-
pretable. Developed by Paatero and Tapper (Paatero and Tap-
per, 1994), PMF is a well-established approach used in previ-
ous source apportionment studies (Polissar et al., 1998; Lee
et al., 1999; Han et al., 2006; Chan et al., 2008; Karanasiou
et al., 2009; Lestari et al., 2009; Santoso et al., 2011; Wimol-
wattanapun et al., 2011). PMF provides more physically re-
alistic results compared to other factor analysis techniques
due to nonnegative constraints in the model and better treat-
ment of missing or below detection limit (BDL) values by
increasing the associated uncertainty (Paterson et al., 1999).
PMF outputs source profiles (F) and source contributions
(G). PMF source profiles were normalized to the percent of
species sum, defined as the percent concentration of an ele-
ment apportioned to a source. An outlier threshold distance
αwas used to reduce the effect of extremely large data points
and was set at a value of 4.0 to be consistent with other PMF
studies (Lee et al., 1999; Han et al., 2006).
Prior to analysis via PMF, the 28 elements measured via
XRF were filtered based on their Pearson’s Rcorrelation
with the total elemental PM mass per mode in order to im-
prove the interpretability of PMF factors. A minimum Pear-
son’s Rvalue of 0.0 was used, which removed elements that
were negatively correlated with the total elemental PM. From
the 28 elements identified by XRF, 20 elements in the coarse
mode, 22 elements in the fine mode, and 19 elements in the
ultrafine mode were included in the PMF analysis. Com-
paring profiles with and without the correlation-based filter-
ing, there was no significant change in factor interpretation.
This indicates that the removed elements were unnecessary
for improving the PMF results (Liao et al., 2019; Ma et al.,
2019). Tables S1–S3 (Supplement) show the correlation co-
efficients of coarse-, fine-, and ultrafine-mode elements. The
filtering of elements through correlation with total PM per
mode was observed to improve the interpretability of the
PMF outputs and remove the need for the matrix rotation pa-
rameter, Fpeak.
Data screening was performed based on the approach of
Polissar et al. (1998) and Han et al. (2006) to ensure that
no erroneous data points were included in the analysis. BDL
values were replaced by half the detection limit and relative
uncertainties were set to 100 % (Han et al., 2006). Signal-to-
noise ratios were determined and elements with low ratios
(less than 0.2) were excluded from the dataset (Paatero and
Hopke, 2003). Measured elemental concentrations below the
detection limit of XRF were replaced with half the detec-
tion limit and their relative uncertainties were set to 100 %
as done in Han et al. (2006). Detection limit values and er-
ror values were based on values provided by the Lawrence
Berkeley National Laboratory.
The current study employs a size-resolved PMF approach
as a supplement to the other analysis methods. PMF is a pow-
erful tool that quantifies the contributions of PM sources and
is useful for forming an initial understanding of the possible
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1260 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
sources from the data. However, PMF may neglect important
events, particularly short-term ones, that can reveal insight-
ful interactions between identified sources and is unable to
dissociate covarying sources as it assumes orthogonality be-
tween factors (Van Pinxteren et al., 2016).
For this study, we included only the DRUM elemental data
for PMF analysis. Speciated data from the PM2.5filter was
excluded due to the limited number of filters available (eight
quartz and eight Teflon filters). The much higher temporal
resolution (174 timestamps) from the DRUM sampler, in ad-
dition to its collection across eight size ranges, provided the
necessary data resolution for PMF while offering the addi-
tional degree of freedom of size-resolved collection. Due to
the limitations inherent in a 2-week research cruise, the col-
lected dataset is not expected to provide a full quantitative
inventory of sources but instead provides an opportunity to
study short-term aerosols events to gain a better understand-
ing of source variability in the SCS region.
3 Results I: mass distributions and time series of
selected elements
3.1 Reconstructed mass and DRUM mass distributions
Mass reconstruction performed on the PM2.5filters shows an
increasing trend in aerosol loadings towards the end of the
cruise (Fig. S1a). A large event beginning on 28 Septem-
ber is characterized by heightened contributions of partic-
ulate organic matter. A smaller aerosol event was also de-
tected by the 23 and 25 September filters. The mass re-
construction shows that 53 % of the total PM2.5gravimet-
ric mass is accounted for by the reconstructed components,
which includes organic carbon (Fig. S1b). The elemental
contribution to the total PM2.5mass was estimated as the
summed contributions of the reconstructed sulfate, sea salt,
and soil components according to formulas from Malm and
Hand (2007) and Chow et al. (2015). Reconstructed elemen-
tal components derived from the DRUM sampler compose
21.2 % of the total PM2.5mass. This is approximately twice
the value calculated with filter-collected elemental concen-
trations (11.7 %). PM2.5Teflon filters have been observed to
show lower concentrations than rotating-drum impactors for
several elements, attributed to insufficient background sub-
tractions when computing for filter concentrations (Venecek
et al., 2016). Other potential factors in this discrepancy in-
clude a complicated sampling environment that may result
in filter losses during collection and the long filter collection
times during the cruise.
Elemental mass size distributions show normalized
species concentrations (dM/d log Dp) across all eight DRUM
stages and can be used to validate the signal of a mode-
specific tracer. In addition to isolating the signal of a tracer,
changes in the mass distributions of key elements over time
indicate periods when mode-specific sources are present.
Figure 3. Time evolution of mass size distributions over the cruise
period: (a) sum of all measured elements, (b) potassium, (c) sulfur,
(d) silicon, (e) vanadium, (f) nickel, (g) iron, and (h) chlorine. The
time periods are colored as follows: 18–19 September (red), 19–24
September (blue), 24–27 September (green), and 27–30 September
(black). Stage numbers are shown in (a).
Figure 3 depicts the mass size distributions of the (a) summed
elemental PM and the following key elements (b) potassium
(K) as a tracer for biomass burning in the fine and ultra-
fine modes; (c) sulfur (S), a general indicator of combustion;
(d) silicon (Si) for soil dust; (e) vanadium (V) and (f) nickel
(Ni), which are often paired as tracers of oil combustion;
(g) iron (Fe), another key tracer for dust; and (h) chlorine
(Cl), a reasonable tracer for sea spray given the sampling
location. Figure 3 is further divided into time periods, dis-
tinguished by color: 18–19 September (red), 19–24 Septem-
ber (blue), 24–27 September (green), and 27–30 September
(black).
The mass distribution of summed elemental PM (Fig. 3a)
is informative, as it shows distinct peaks in the coarse and
sub-micrometer ranges, pointing to a combustion or anthro-
pogenic signal during the cruise. The total mass size distribu-
tion shows that, over time, a regime change occurred around
24 September, during which the general back trajectory ori-
gin shifts to the Maritime Continent. Comparing the mag-
nitude of the summed mass distribution to those of the key
species, it is clear that S contributed a significant part of the
sub-micrometer mass. Elements associated with combustion
showed peaks in stage 5 (0.56–0.75 µm) and stage 7 (0.26–
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1261
0.34 µm). K, S, and Si have very similar changes in their
mass size distributions over the cruise period that are sug-
gestive of a common source (Fig. 3b–d). During the latter
half of the cruise, a regime shift occurred, wherein back tra-
jectory origins shifted to southern Kalimantan (Fig. 2). We
observe coincident enhancements in K, S, and Si – indica-
tive of a common source, likely biomass burning. These ele-
ments have strong peaks in stages 5 and 7 during the whole
cruise, but particularly high values are observed during the
last days of the sampling period (27–30 September). A gen-
eral enhancement late in the cruise is likely related to the in-
crease in the number of active fire hotspots reported by Reid
et al. (2015), who attributed these hotspots primarily to In-
donesian Kalimantan and southern Sumatra. As the cruise
took place during the end of the Asian summer monsoon,
300 m a.g.l. winds were predominantly southwesterly. A shift
in back trajectories at the end of the cruise to the western and
southern coasts of Borneo is observable in Fig. 2l, suggest-
ing the source of the late-cruise enhancement to be the MC,
which hosts elevated aerosol background levels during this
time of year from seasonal burning (Reid et al., 2013). The
advection of this large aerosol event can be observed in the
NAAPS smoke model over the region (Fig. 2g, h). The at-
tribution of late-cruise aerosol enhancement to the MC is in
agreement with Reid et al. (2015), who noted that the AOD
maps and southwesterly flows towards the end of the cruise
were suggestive of southwesterly transport from the MC to
SCS.
Covariance of Si (Fig. 3d) with K and S suggests possible
fine soil entrainment caught in burning updraft (Reid et al.,
2015). The stage 5 and stage 7 peaks in S are similar to those
observed for northern SCS in the springtime (Atwood et al.,
2013a); however, we report enhanced values, attributed to the
timing of the sampling period during the MC burning season.
Interestingly, Si shows a strong peak early in the cruise
(18–19 September) unique to the ultrafine mode, which indi-
cates this particular signal may not originate from soil dust
but from fly ash (Xie et al., 2009). As the Vasco was traveling
past the islands of Mindoro and Coron en route to Palawan,
local sources are likely the cause of the ultrafine Si enhance-
ment. This early cruise Si signal is further examined through
later time series and regressions.
V shows a mass distribution characteristic of a combus-
tion source with strong peaks in stage 5, stage 7, and stage
8 (0.10–0.26 µm) (Fig. 3e). Almost no contribution was ob-
served for coarser stages 1 through 4 (0.75–10 µm), indicat-
ing that V did not originate from soil (Lin et al., 2015) and
can be treated as a tracer for oil combustion. Ni shows a sim-
ilar mass distribution (Fig. 3f) but had a larger spread over
the eight stages than V, which may be due to contributions
from other sources such as fly ash (Davison et al., 1974).
Fe and Cl, well-known tracers for soil dust and sea spray,
respectively, showed coarse-mode distributions that taper off
considerably in the sub-micrometer stages (Fig. 3g, h). Cl
shows a purely coarse distribution, indicative of the influence
Table 1. PM1.15 /PM10 ratio slopes for elements ordered by ratio
slope.
Ratio R2Ratio Standard
slope correlation average deviation
V 0.94 0.99 0.95 0.07
K 0.82 0.94 0.35 0.21
S 0.8 0.92 0.49 0.17
Zn 0.74 0.94 0.62 0.04
Y 0.7 0.7 0.53 0.11
Zr 0.7 0.63 0.65 0.07
Mo 0.7 0.67 0.65 0.04
Ti 0.68 0.7 0.53 0.08
Rb 0.61 0.64 0.73 0.09
Al 0.51 0.68 0.55 0.12
Pb 0.47 0.44 0.67 0.06
Cu 0.4 0.42 0.63 0.05
Ni 0.31 0.33 0.61 0.08
As 0.31 0.36 0.33 0.26
Mn 0.3 0.62 0.49 0.19
Si 0.29 0.56 0.32 0.13
Se 0.2 0.24 0.59 0.06
P 0.19 0.32 0.27 0.08
Na 0.16 0.57 0.17 0.03
Sr 0.16 0.11 0.49 0.08
Br 0.13 0.17 0.47 0.08
Ca 0.07 0.59 0.1 0.05
Cl 0.06 0.67 0.04 0.02
Fe 0.06 0.38 0.24 0.12
Mg 0.03 0.29 0.07 0.03
Co 0.03 0.03 0.57 0.1
Ga 0.03 0.04 0.56 0.09
Cr 0.01 0.02 0.19 0.19
of sea spray considering the sample location (Viana et al.,
2008; Gugamsetty et al., 2012; Farao et al., 2014). Fe shows
small peaks in stage 4 (0.75–1.15 µm), stage 5, and stage 7;
however, these do not constitute a significant signal relative
to its coarse-mode concentrations. As such, we treat Fe as our
coarse-mode soil dust tracer. The mass distribution of Fe is
observed to increase across stages 1 through 3 (2.5–10 µm)
over the cruise period. The increase in coarse Fe coincides
with the NAAPS-simulated transport of smoke (Fig. 2g, h)
and mirrors the enhancements of K, S, Si (Fig. 4a, b), and Al
(Fig. S2a, b). These patterns suggest that coarse soil dust ac-
companies smoke emissions, possibly through entrainment.
The presence of soil dust is further corroborated by Fig. 3d,
which shows the presence of Si in the coarse mode. The dis-
tinct coarse- and fine-mode peaks of Al and Si indicate sep-
arate soil dust sources. As fine-mode particles have longer
residence times (Cohen et al., 2010a), the fine peaks may be
an indicator of long-range transport of fine soil dust through
the SCS.
Interpreting DRUM data reveals insights about the com-
position and interpretation of sources. Table 1 shows the ra-
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1262 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
Figure 4. Time series of (a) stage 4–6 K, S, and Si; (b) stage 7–8 K, S, and Si; (c) stage 4–6 V and Ni; (d) stage 7–8 V and Ni; and (e) stage
1–3 Fe and Cl.
tios of elemental PM1.15 /PM10 mass concentrations. As in
Atwood et al. (2013a), the ratio slope was computed by tak-
ing the slope of the linear regression line between elemen-
tal PM1.15 and PM10 mass concentrations, accompanied by
r2values. Direct averages of per-timestamp ratios of PM1.15
and PM10 were also taken to compute for ratio averages, ac-
companied by the standard deviation of the ratios. Fe and Cl
both had ratios of 0.06, which confirms the predominantly
coarse nature of these species. As commonly used tracers
of soil dust, Al and Si show moderate ratio slope values
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1263
of 0.51 and 0.29, respectively, suggesting that Al resides in
both coarse and fine (PM1.15) modes, while Si is predomi-
nantly coarse. As expected, elements commonly associated
with anthropogenic species such as V, K, and S show high
PM1.15 /PM10 ratios (0.8 and above), which indicates that
these elemental particles largely reside in the fine and ultra-
fine modes. The high ratios of V, K, and S provide evidence
for the presence of anthropogenic emissions from sources
such as oil combustion and biomass burning, while the low
ratios of Fe and Cl support their treatment as tracers for soil
dust and sea spray, respectively.
The time-resolved DRUM data are important for showing
variations in species which may be representative of impor-
tant aerosol events. Thus, observations on the time-resolved
DRUM data can aid in our analysis. At the beginning of the
cruise, between 18 and 19 September, V, Ni, and Si show en-
hancements in stages 5, 7, and 8. The stage 7 Si peak during
this time is the maximum concentration over the entire cruise
period, therefore this warrants further analysis through later
time series and regressions. The period of 19–24 September
shows a low point in the DRUM peaks of several elements,
most notably combustion tracers K and V (Fig. 3b, e), while
Cl (Fig. 3h) shows higher peaks in the coarse mode, which
suggests a period of clean marine aerosol. This period was
described by Reid et al. (2015) as the cleanest of the cruise.
The NAAPS model shows nearly zero smoke concentration
at the sampling site (Fig. 2f), while 72 h HYSPLIT back tra-
jectories indicate that air masses originate from the central
SCS (Fig. 2j). From 24 to 27 September, we observed the
first major aerosol event, characterized by the stage 5 and
7 enhancements of several combustion elements: K, S, Si,
V, and Ni (Fig. 3b–f). Fe, our coarse-mode soil dust tracer,
shows enhancements in stages 1 to 3 (Fig. 3g), which points
to combustion-related entrainment of soil dust in the coarse
mode. The NAAPS model (Fig. 2g) depicts the intensifica-
tion and spread of a smoke-related aerosol event that had
been escalating in southern Kalimantan since 22 Septem-
ber, reaching the Vasco around 26 September. During this
mid-cruise period, concentrations of biomass burning species
K, S, Si, and Al are elevated and oil combustion tracers V
and Ni show their maximum concentrations for the cruise in
stages 5 and 7 (Fig. 4e, f). The last period, 28 to 30 Septem-
ber, depicts the highest concentrations of elements associ-
ated with biomass burning (Fig. 4a, b; Fig. S2a, b). As seen
in the NAAPS smoke model (Fig. 2h) and HYSPLIT model
(Fig. 2l), the westward movement of TC Nesat across the re-
gion alters back trajectories so that they travel around Bor-
neo island, reaching southern Kalimantan, which hosted a
high active fire hotspot density during this time (Reid et al.,
2015), thus bringing polluted air masses toward the sampling
site. Stage 5 and 7 peaks of K and S are quite notable as no
other stages show significant enhancements in response to
this event. Fe and Si show similar changes but instead for the
coarser stages 1 to 3 (Fig. 3d, g), indicating a covariance of
soil dust and biomass burning tracers. The temporal trends
from the DRUM data serve as an entry point into the time
series analysis. By identifying key DRUM stages and time
periods per element based on their mass size distributions,
we can then examine these stages to observe aerosol events
over the cruise period.
3.2 Time series of selected elements
The first few days of the cruise showed an 18 September
event in oil combustion tracers V and Ni in the ultrafine mode
(Fig. 4d), with a coincident but lower-magnitude response in
the fine mode (Fig. 4c). Ultrafine-mode V and Ni show their
maxima for the cruise period during this time, expanded fur-
ther in Sect. 5. High concentrations of ultrafine Si were sam-
pled during this time from the beginning of the cruise until
19 September when it dropped to stable background levels.
This early cruise enhancement was also seen in its mass dis-
tribution plot (Fig. 3d). As the Vasco was traveling among
islands, the Si signal may be due to local sources en route to
the El Nido sampling site.
Reid et al. (2015) noted periods of clean regime after de-
parting Manila Bay through midday 22 September, observ-
able in the consistently low concentrations of various ele-
ments (Fig. 4). Chlorine shows a gradual increase in con-
centration from 20 September until 24 September. Chlorine,
although it ages into HCl, is assumed to be fresh due to the
sampling location and can therefore be used as an indicator
of sea spray. Interestingly, coarse-mode Cl (Fig. 4e) showed
peak concentration times during low points in the concen-
trations of anthropogenic aerosol species (Fig. 4a–d), mark-
ing periods of clean marine aerosol on 22–24 September
and 26–28 September. Wet deposition processes are likely
responsible for the suppressed anthropogenic aerosol con-
centrations as precipitation was prevalent during these pe-
riods (Reid et al., 2015). Conversely, peaks in the concen-
trations of anthropogenic aerosol occurred during dry peri-
ods of the cruise when precipitation was low: 24–26 Septem-
ber and 28–30 September. During the periods of clean ma-
rine aerosol, back trajectories shift away from source regions
and traverse open sea (Fig. 2j, k), which also hosts a lower
shipping route density compared to coastal regions (Fig. S3,
Supplement). The first half of the cruise also saw the lowest
concentrations from species associated with biomass burn-
ing, specifically sub-micrometer K, S, Si, (Fig. 4a, b), and
Al (Fig. S2a, b, Supplement). These species track each other
quite well throughout the cruise period, indicating a common
source.
The event between 24 and 26 September is observable
on the time series of several key elements. The plume was
the first of two distinct plume events reported by Reid et
al. (2015), with the later plume occurring on 29 Septem-
ber. The enhancement of all elements in Fig. 4 suggests a
mix of biomass burning, oil combustion, and soil dust in-
fluences within the 24–26 September plume. Fine-mode V
and Ni show their maximum concentrations for the cruise
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1264 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
during this event (Fig. 4c). Although these two plumes ap-
peared as one uniform progression across the SCS region on
the NAAPS smoke model (Fig. 2h), the time series showed
the presence of two distinct events (Fig. 4), which is corrob-
orated by observations from Reid et al. (2015). During this
period, aerosol concentrations dropped sharply before recov-
ering due to the passage of squall lines, observed in the time
series for K, S, Si, Fe, and Cl (Fig. 4a, b, e). As concluded
in Reid et al. (2015), frequent short-term events, such as cold
pools and squall lines, must be accounted for in modeling
studies in order to properly capture aerosol–convection inter-
action.
The period between plumes (26–28 September) is char-
acterized by an overall drop in the aerosol concentration of
species associated with anthropogenic sources (K, S, V, and
Ni; Fig. 4a–d). As Cl concentrations show peak values dur-
ing this period (Fig. 4d), this indicates a period of pure ma-
rine aerosol sampling similar to the 22–24 September clean
period. Coinciding with the passage of TC Nesat through the
SCS, the observed drop in aerosol concentration is attributed
to a possible restriction of shipping traffic in response to the
TC and scavenging of aerosols by precipitation along the TC
inflow arm (Fig. 2c) (Reid et al., 2015).
The last days of the cruise were particularly eventful, as
the largest aerosol event of the cruise period was visible on
the NAAPS model in the form of smoke (Fig. 2h), accompa-
nied by the spread of high AOD values throughout the SCS
(Fig. 2d). Although the large areas of cloud cover created
by TC Nesat hinder the detection of AOD on 26 September,
the region was free enough of cloud cover by 29 Septem-
ber that significant AOD values were observed to visibly
stretch from southern Kalimantan towards the Vasco sam-
pling site (Fig. 2d). In general, the NAAPS smoke transport
model agrees with the spatial distribution of high AOD. Here,
NAAPS modeling of smoke transport is useful in demon-
strating the event’s northward advection and the severity of
smoke concentration on Borneo on 26 September (Fig. 2h).
Time series plots of elements associated with biomass burn-
ing (K, S, Si; Fig. 4a, b) and coarse-mode soil dust (Fe;
Fig. 4d) show significant enhancements during this time that
were also observed on their mass distributions (Fig. 3). HYS-
PLIT back trajectories show that air masses originate from
southern Kalimantan during this period, as opposed to Penin-
sular Malaysia during the first half of the period (Fig. 2j, l).
The shift in air mass trajectories is attributed to the passage
of TC Nesat through the region as inflow arms from TCs
have been observed to accelerate air mass advection across
the SCS, bringing more MC air into the region (Reid et al.,
2012, 2015). The observed transport of emissions from Bor-
neo indicates that TC-enhanced long-range transport is a sig-
nificant factor in SCS aerosol dispersion.
4 Results II: positive matrix factorization and
regressions
4.1 Source apportionment via positive matrix
factorization
To verify groupings of key elements and aid in source
identification, size-resolved PMF was performed. As de-
scribed in Sect. 2, the eight-stage DRUM data were com-
bined into coarse (1.15–10 µm), fine (0.34–1.15 µm), and ul-
trafine (0.10–0.34 µm) modes and the species included in the
PMF analysis were then filtered based on their correlation to
the aggregated PM concentration per mode. The PMF analy-
sis resolved six sources across the three size ranges: biomass
burning, oil combustion, soil dust, a crustal–marine mixed
source, sea spray, and fly ash (Table 2). Due to the similari-
ties in composition and temporal trends of the crustal–marine
mixed source in the coarse mode and the sea spray factor in
the fine mode, they are depicted together in Figs. 5–7 for
simplicity.
One strength of PMF is its quantification of a source’s con-
tribution. Figure 5 shows the percent contribution of each
source relative to the total elemental PM mass. As expected,
natural sources such as the crustal–marine mixed source
and soil dust mainly contribute to the coarse mode, while
combustion-related sources such as biomass burning and oil
combustion contribute to the fine and ultrafine modes. The
identification of sea spray in the fine mode is likely due to the
existence of Cl in stage 4 of the DRUM sampler (Fig. 3h).
The existence of these sources in their expected modes is
an indicator of the successful implementation of PMF. The
following sections describe the observed characteristics of
sources determined by PMF.
4.1.1 Crustal–marine mixed source
The crustal–marine mixed source was resolved in the coarse
mode and is characterized by high apportionments for Mg,
Cl, P, Al, Si, S, and Ca (Fig. 6a). This source explains nearly
half of the variation in crustal elements such as Al, Si, and
Ca. Na and Cl show the highest contribution to the fac-
tor mass, which indicates marine influence (Fig. S4, Sup-
plement). These elements are indicative of a mix of marine
and crustal emissions (Han et al., 2006; Wang et al., 2014),
thus its identification as a crustal–marine mixed source. The
mixed nature of the source points to the covariance of lo-
cal crustal emissions from islands of the Maritime Conti-
nent and those nearby with sea spray. Cl has been treated
as the tracer for this factor due to its high factor sum appor-
tionment (Fig. S4, Supplement) and is considered marine in
origin under the assumption that the sampled Cl originated
from freshly produced sea spray (Atwood et al., 2013a). This
is likely the case for the cruise, as sampling was done over
sea water. The factor showed quite high mass contributions
to the coarse mode (56.8 %) indicating its dominant influ-
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1265
Figure 5. Contributions of factors to the total elemental PM mass. The crustal–marine mixed source (coarse mode) and sea spray (fine mode)
share the same color to reflect their similar chemical compositions.
Figure 6. PMF source profiles across different size ranges displayed by percent of species sum for (a) crustal–marine mixed source, (b) soil
dust, (c) biomass burning, (d) oil combustion, and (e) fly ash. Coarse: stage 1–3 (1.15–10 µm; blue); fine: stage 4–6 (0.34–1.15 µm; orange);
ultrafine: stage 7–8 (0.10–0.34 µm; green).
ence on coarse elemental PM (Fig. 5a). Although both this
factor and the coarse-mode soil dust factor are related to
crustal emissions, the crustal–marine mixed source is distinct
from the coarse-mode soil dust factor in terms of its tempo-
ral trend, most apparent during the 28–30 September aerosol
event (Fig. 7a, b).
4.1.2 Sea spray
This factor was resolved in the fine mode and shows high
apportionments for Na, Mg, Cl, and Ca. The identification
of the factor as sea spray is evidenced by the nearly 100 %
source apportionment of Cl. This factor showed fine (30.4 %)
modes, attributed to the sampling location over water. As
noted above, the appearance of this factor in the PMF analy-
sis is due to the persistence of Cl in the 0.75–1.15 µm of the
DRUM sampler (Fig. 3h). The covariance of the sea spray
factor in the fine mode with the crustal–marine mixed source
in the coarse mode points to the influence of marine emis-
sions to some extent in both the fine and coarse modes, as
suggested by a moderate correlation coefficient (0.67) be-
tween PM10 and PM2.5Cl (Table 1).
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1266 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
Table 2. Sources identified in each size range with PMF: coarse (1.15–10 µm), fine (0.34–1.15µm), and ultrafine (0.10–0.34µm).
Source Major components Coarse Fine Ultrafine
Biomass burning K, S, Si, Al, As + +
Oil combustion V + +
Crustal–marine mixed source Mg, Cl, P, Al, Si, S, Ca +
Sea spray Na, Mg, Cl, Ca +
Soil dust Fe, Al, Si, Ca, Ti, Zn + +
Fly ash As, Se, Pb, Zn, Ti + + +
4.1.3 Soil dust
This factor was characterized by the presence of Fe, Al, Si,
K, Ca, Ti, and Zn in the coarse mode and Fe, Cr, Mn, and Y
in the fine mode (Fig. 6b; Table 2). Several of these elements
are associated with soil dust (Artaxo and Maenhaut, 1990;
Artaxo et al., 1998; Lestari et al., 2009; Wimolwattanapun et
al., 2011; Gugamsetty et al., 2012). Soil dust may originate
from the nearby island of Palawan but can also potentially
come from Borneo. The PMF model was able to distinguish
between the crustal–marine mixed source and soil dust fac-
tors. As crustal–marine mixed emissions are assumed to be
freshly sampled during the cruise and the temporal trends
of the two sources are distinct (Fig. 7a, b), this suggests the
possibility of a long-range transport mechanism for coarse-
mode soil dust. The time series of coarse soil dust (Fig. 7b)
tracks the fine biomass burning factor well (Fig. 7c), indica-
tive of coarse soil dust particles entrained in biomass burning
plumes. Fe serves as our tracer for soil dust due to its high
apportionment in both soil dust modes. This factor showed
mass contributions of 18.7 % and 14.0 % in the coarse and
fine modes, respectively, which indicates the predominantly
coarse-mode contribution of the factor (Fig. 5a, b).
4.1.4 Biomass burning
This factor was characterized by high levels of K and S and
moderate levels of Al, As, and Si that were found to be as-
sociated with biomass burning in previous studies (Artaxo et
al., 1998; Han et al., 2006; Lestari et al., 2009; Atwood et
al., 2013a; Alam et al., 2014) (Fig. 6c; Table 2). This fac-
tor showed the highest percent contributions to the PM mass:
32.4 % and 45.9 % in the fine and ultrafine modes, respec-
tively. The sources of the 26 September and 28–30 Septem-
ber events (Fig. 7c) will be investigated in Sect. 5. The pres-
ence of crustal elements Fe, Si, and Al in the source profile
and the covariance of the coarse soil dust factor (Fig. 7b)
with this factor (Fig. 7c) indicate possible soil dust entrain-
ment during burning updraft (Reid et al., 2015; Schlosser et
al., 2017).
Figure 7. Temporal distribution of PMF source contributions
(µg m−3) for (a) crustal–marine mixed source, (b) soil dust,
(c) biomass burning, (d) oil combustion, and (e) fly ash. Coarse:
stage 1–3 (1.15–10 µm; blue); fine: stage 4–6 (0.34–1.15 µm; or-
ange); ultrafine: stage 7–8 (0.10–0.34 µm; green).
4.1.5 Oil combustion
This factor was characterized by high levels of V (Fig. 7d;
Table 2), a well-documented tracer for oil combustion (Hed-
berg et al., 2005; Mazzei et al., 2008; Becagli et al., 2012).
As shown in Fig. 5, the oil combustion factor only appeared
in the fine and ultrafine sizes, contributing 7.3 % and 26.1 %,
respectively, to the total elemental PM mass. The increasing
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1267
contribution towards finer stages corroborates the identifica-
tion of the factor as an anthropogenic source. The presence
of oil combustion is expected as the SCS hosts high shipping
volume, particularly in parts of the Borneo coast (Fig. S3,
Supplement).
4.1.6 Fly ash
This factor was observed in all size modes, characterized
by high levels of trace metals Ni, Ga, Zn, Se, Br, Rb, and
Pb across modes, with slight differences in composition per
mode (Fig. 6e) and a source contribution without distinct
events (Fig. 7e). The dominance of Ni, Zn, Se, and Br are
indicative of fly ash (Davison et al., 1974; Markowski et al.,
1985; Deonarine et al., 2015). Moderate apportionments of
crustal elements Na, Mg, Al, Si, P, and Ti are also observed,
suggestive of entrained soil. The source contribution time se-
ries shows a background-type signal. The factor contributed
24.5 %, 16.0 %, and 28.0 % to the total elemental PM mass
for the coarse, fine, and ultrafine modes, respectively (Fig. 5).
Long-range transport of fly ash from coal-fired power plants
in Indonesia or Peninsular Malaysia may be responsible for
the appearance of the factor, as no local coal-fired power
plants could be found upwind of the sampling site in 2011.
The PMF analysis resolved the presence of six sources
across the ultrafine, fine, and coarse modes, which aids in
directing further analysis by identifying key species in the
source profiles. Pearson correlation heat maps (Fig. S5–S7,
Supplement) and matrices with numerical values (Tables S1–
S3, Supplement) were constructed to examine the relation-
ships between species. The first column of the correlation
outputs (Fig. S5–S7, Tables S1–S3, Supplement) shows the
correlation coefficient of the element when compared to the
summed elemental PM for that mode. Similar groupings of
elements were observed when compared to the PMF source
profiles, indicating the robustness of the analysis. In the
coarse mode (Fig. S5, Table S1, Supplement), we observe
high correlations between Na, Mg, Cl, P, S, K, Ca, Br, and
Sr, which are associated with sea spray and crustal sources
(Han et al., 2006; Wang et al., 2014). Fe, Ti, Mn, Si, and Zn
show moderate to high correlations in the coarse mode, in-
dicative of dust (Karanasiou et al., 2009; Wimolwattanapun
et al., 2011; Lin et al., 2015; Landis et al., 2017). In the fine
mode, moderate to high correlations between Al, Si, P, S, K,
and Br are observed (Fig. S6, Table S2, Supplement). Sev-
eral of these biomass burning elements show similarly strong
correlations in the ultrafine mode (Fig. S7, Table S3, Supple-
ment). V and Ni show a high correlation coefficient (0.91) in
the ultrafine mode, indicative of oil combustion.
The excellent correspondence between the observed
groupings of elements based on correlation (Tables S2–S4,
Supplement) and the sources resolved by PMF (Table 2)
adds confidence to the identification of key sources during
the cruise. However, as PMF is an unsupervised technique,
it may not sufficiently disaggregate significant, consecutive
aerosol events. Visually, two distinct ultrafine events occur
between 18 and 19 September in Si (Fig. 4b) and V and
Ni (Fig. 4d), which are merged by PMF in its oil combus-
tion factor (Fig. 7d). The disproportionate enhancement of
ultrafine-mode Si over V and Ni suggests a source other than
oil combustion. Thus, to further expand on the relationships
between elements, we turn to regression analysis.
4.2 Regressions of selected elements
An ultrafine Si event between 18 and 19 September was
shown in the mass size distribution (Fig. 3d) and the time
series (Fig. 4b) of ultrafine Si. Fly ash was the hypothe-
sized source of the ultrafine Si signal; however, although
the PMF analysis suggested the presence of fly ash, ultra-
fine Si was not significantly apportioned to the fly ash factor
(Fig. 6e). Additionally, none of the factor contributions from
PMF showed a similar trend between 18 and 19 September
to that of ultrafine Si. This suggests that PMF may have mis-
handled the early Si enhancement (Fig. 4b) by merging it
with an enhancement in V and Ni that occurred soon after
(Fig. 4d). Regressions show that between 18 and 19 Septem-
ber Si had distinct ratio slopes and moderate correlations
with P (r2=0.76), S (r2=0.73), and Al (r2=0.61) (Fig. 8;
Table S4, Supplement) but poor correlations with fly ash trac-
ers (As, Se, and Pb; r2<0.12). The high correlations of Si
with P, Al, and S suggest a distinct source of Si between 18
and 19 September versus the rest of the cruise; but the low
correlations with fly ash tracers rule out fly ash as a possible
source. As the Vasco was traveling near islands, the source
of the ultrafine Si enhancement is likely a local source en
route to Palawan. The sudden enhancement may be related
to a rapid nucleation event, as even sub-micrometer dust can
be an important source of CCN in marine and coastal envi-
ronments (Twohy et al., 2009).
As S is an indicator of general combustion (Atwood et al.,
2013a), it is important to elucidate its relationship with trac-
ers of other combustion sources. Multiple linear regression
was performed on S on the fine and ultrafine modes (Fig. S8,
Supplement). It was found that K and V were excellent pre-
dictors of S for most of the cruise but the model required
the addition of Al to capture the variance in S between 24
and 26 September, suggesting an additional source during
this period separate from biomass burning or oil combus-
tion. A detailed description of the multiple linear regression
analysis can be found in the Supplement. Further examining
the relationships of S to these combustion sources, fine- and
ultrafine-mode linear regressions of K and V, colored by the
concentration of S per given time, were constructed to show
the relationships between the three species (Fig. 9a, b). S co-
varies more with K than V, as seen with the clearer color gra-
dient along the K axis, suggesting the origin of S during the
cruise to be more dominantly from biomass burning rather
than oil combustion.
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1268 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
Figure 8. Linear regressions of ultrafine Si and its most highly correlated elements, (a) P, (b) S, and (c) Al, divided into the cruise periods
before 19 September (red) and after 19 September (blue).
The ratio between V and Ni is often used as an indicator
of the type of oil combustion source (Hedberg et al., 2005;
Nigam et al., 2006; Mazzei et al., 2008; Becagli et al., 2012;
Lin et al., 2015). Linear regression plots of V and Ni have
a slope of 3.64 in the ultrafine mode (Fig. 9c). Nigam et
al. (2006) measured a V /Ni ratio of 3.5–4 when sampling
shipping emissions directly from the exhausts of various ship
engines, which suggests shipping to be the main source of
ultrafine-mode oil combustion during the cruise.
As soil composition varies geographically, soil dust ra-
tios are excellent indicators of a plume’s origin (Prospero
et al., 1999; Song et al., 2006; Witt et al., 2006). Figure 8d
shows linear regressions of soil dust elements in the coarse
and fine modes. Al and Si, well-known indicators of dust
(Viana et al., 2008; Tian et al., 2016; Landis et al., 2017),
show moderate correlations with each other in the coarse
and fine modes but slightly differ in ratio slopes between
the fine (Al /Si ∼1.3; r2=0.94) and coarse (Al/Si ∼0.93;
r2=0.78) modes (Fig. 9d). This is indicative of varying
sources of fine- and coarse-mode soil, with coarse-mode soil
dust enriched in Si; however, this could also be a matrix
effect from the XRF analysis. As the Vasco remained near
Palawan island, local dust could be the source of coarse-
mode Si enrichment; however, soil dust from Borneo is also
a possibility.
The regression analysis showed an early cruise enhance-
ment in ultrafine Si that was merged by PMF with a V and Ni
enhancement that occurred soon after, highlighting the im-
portance of the regression analysis in addition to PMF to in-
vestigate the temporal characteristics of sources via elemen-
tal tracers. We suggest a local source en route to the main
sampling area to be the cause of the enhancement, but fly ash
is unlikely to be the source due to low correlations with its
tracers As, Pb, and Se. The analysis also showed the strong
associations of S with biomass burning and oil combustion;
however, S was shown to covary more significantly with the
former. Oil combustion was determined to originate from
shipping, as indicated by a V /Ni ratio within the range of
that measured by a previous shipping emission study. Finally,
we infer multiple sources of soil dust between the coarse and
fine modes due to distinct Si–Al ratios between modes; how-
ever, we are unable to determine the exact sources due to lack
of information regarding local and regional soil dust ratios.
5 Results III: back trajectory analysis
5.1 18–19 September: ultrafine V and Ni enhancement
from Sandakan, Sabah
As described in Sect. 3, ultrafine-mode V and Ni show a
maximum around 18 September (Fig. 4d). As the Vasco was
traveling near local islands, the event may originate from a
local source; however, back trajectories propose an oil com-
bustion source on Borneo. Back trajectories were generated
every hour between 14:00 and 18:00 UTC (corresponding to
22:00 and 02:00 LT) on 18 September and show a westward
shift along the eastern coast of Borneo (Fig. 10a). The coast
of Borneo is largely forest (Fig. 10b) but hosts the city of
Sandakan, one of Sabah’s major ports (Fig. 10c, d). In ad-
dition to shipping traffic (Fig. 10d), Sandakan contains oil
depots that are a major source of industry in the area. During
the westward shift of the back trajectories, air masses pass
through Sandakan at around 16:00 UTC, approximately the
time of the sampled spike in V. The shipping activity and oil
depots present in this area may be responsible for the spike in
oil combustion tracers, indicating the complexity of aerosol
transport in the region, as small cities like Sandakan may be
a source of significant spikes in aerosol.
5.2 20–24 September: clean marine period
The first half of the cruise showed the lowest concentrations
of elements associated with biomass burning K, S, Si, and
Al. Back trajectories during this early period originate from
the northern part of Borneo and do not penetrate deeply into
the MC until late into the cruise (Fig. 2l). During this period,
HYSPLIT back trajectories show that air mass pathways shift
away from the Borneo coast towards open sea (Fig. 2j). In
addition to the shift away from biomass burning sites, back
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1269
Figure 9. Scatterplot of key species during the cruise. (a) Fine-mode K and V colored by the concentrations of S at a given time, (b) ultrafine-
mode K and V likewise colored by concentrations of S at a given time, (c) ultrafine-mode V and Ni, and (d) coarse- and fine-mode Al and
Si.
Figure 10. Determination of the 18 September event using (a) HYSPLIT back trajectories, (b, c) a Google Maps view of the northeastern
coast of Borneo (map data ©2018 Google), and (d) the density of shipping traffic from Sandakan, Sabah (source: MarineTraffic). Red squares
indicate the location of the succeeding plot.
trajectories between 22 and 24 September pass through areas
of open sea that host lower levels of shipping traffic (Fig. S3,
Supplement).
5.3 24–26 September: large mixed aerosol event from
northwestern Borneo
Around 26 September, increases in fine-mode V and Ni oc-
curred when air masses passed through the northwestern
coast of Borneo, suggesting the presence of ports or oil de-
pots, similar to the aforementioned spike on 18 Septem-
ber from Sandakan. Back trajectories generated every 6 h
starting from 24 September 15:00 UTC until 26 September
09:00 UTC show little change over this period (not shown)
and intersect with the shipping route hub located along north-
western Borneo, which would explain the V and Ni spikes
(Figs. 2k, S1, Supplement). The enrichments of biomass
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1270 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
burning and combustion tracers K and S in the sampled air
mass span a wider period from 24 September to 26 Septem-
ber. This may be due to burning activity along the coast
of Borneo, which hosts several MODIS-detected active fire
hotspots. Late-night land breeze from the island may have
advected polluted air masses towards the coast.
5.4 28–30 September: large biomass burning event
from southern Kalimantan
Enhancements of these elements after 28 September coincide
with a regional increase in AOD (Fig. 2d) and are captured
by the NAAPS model in the form of a large smoke event ad-
vected northeast (Fig. 2h). Linear regressions show this large
aerosol event at the end of the cruise as a distinct group of
points with enhanced concentrations of K and S (Fig. S9,
Supplement), suggesting an increase in biomass burning ac-
tivity during this time. Reid et al. (2015) observed a sharp
increase in the number of active fire hotspots, particularly in
Sumatra and southern Kalimantan. As discussed above and
depicted in Fig. 2, TC Nesat played a major in role in syn-
optic wind patterns during the cruise, causing a shift in back
trajectories after 28 September to the southwest coast of Bor-
neo. Thus, the enhancements of sub-micrometer K, S, Si, and
Al likely originate from biomass burning in the MC.
6 Summary and conclusions
This study describes the size-resolved aerosol elemental
composition of particles collected by a DRUM rotating im-
pactor during the 17 to 30 September 2011 M/Y Vasco
cruise in the vicinity of Palawan island in the Philippines.
This region was chosen due to its location as a receptor
for MC aerosol sources, such as biomass burning, oil com-
bustion, and soil dust. Meteorological conditions during the
cruise were conducive to southwesterly long-range trans-
port for seasonal burning aerosol, which was observed in
the concentration time series of tracers and satellite-derived
AOD. Size-resolved aerosol composition in the coarse (1.15–
10 µm), fine (0.34–1.15 µm), and ultrafine (0.10–0.34 µm)
modes were used as key tracers to ascertain source contri-
butions. Despite the meteorological complexity of the SCS,
we can gain insights into aerosol sources by focusing on key
elemental species. The time series of key elements showed
distinct events on 18–19, 24–26, and 28–30 September, with
clean aerosol periods between events. These aerosol events
served as case studies of sources in the region. While biomass
burning is indeed a key source of aerosol, other sources such
as oil combustion, crustal–marine mixed source, fly ash, and
soil dust contribute to the chemical profile of the SCS dur-
ing the southwest monsoon. Understanding these sources is
key to characterizing aerosol composition and transport in
the SCS and, by extension, developing our understanding
of aerosol–cloud behavior in the region. As back trajectory
analysis and aerosol chemistry showed the presence of mul-
tiple key sources, the general conclusions of the study are as
follows.
1. Mass distributions of key elements showed the evolu-
tion of aerosol chemistry throughout the cruise and in-
teresting covariances between modes. Stage 5 (0.56–
0.75 µm) and stage 7 (0.26–0.34 µm) showed enhanced
peaks in several elements associated with combustion.
Throughout the cruise, mass distributions of V and
Ni track each other well, both temporally and across
DRUM stages, indicative of oil combustion. Mass dis-
tributions of V and Ni show higher values in the ul-
trafine mode between 18 and 19 September, indica-
tive of an early oil combustion-enriched air mass which
was identified to possibly originate from Sandakan,
Sabah, Borneo. Mass distributions of K, Al, and S show
large enhancements in the fine and ultrafine modes after
27 September, corroborated by a reported large aerosol
event from Reid et al. (2015). The strong peaks of these
biomass burning tracers, in combination with the rapid
spread of high AOD and NAAPS-modeled smoke con-
centration across the region, provide evidence for inten-
sive emissions from the MC. Coarse-mode soil dust el-
ements such as Fe and Si showed similarly timed en-
hancements, attributed to soil particle entrainment dur-
ing burning.
2. Short-term meteorological events such as TC Nesat
played a key role in long-range transport as they propa-
gated through the region, expediting the northeastward
advection of aerosol emissions, an effect observed in
previous studies (Atwood et al., 2013a; Reid et al.,
2012, 2015). The sudden variations in aerosol concen-
tration after 24 September can be connected to the
movement of TC Nesat through the region. Prior to
these events, aerosol concentrations remained at gen-
erally low levels, as NAAPS shows smoke was largely
constrained to the Southern Hemisphere. The passage
of TC Nesat advected air masses more northward, al-
lowing them to penetrate deep enough into the North-
ern Hemisphere to be sampled by the Vasco. The TC’s
passage coincided with a shift in air mass origin from
Peninsular Malaysia prior to 24 September to areas
known for intense burning activity, most notably south-
ern Kalimantan, by the end of the cruise. This corre-
sponded to a mixed aerosol event from 24 to 26 Septem-
ber attributed to Brunei, Borneo, and a significant in-
crease in biomass burning tracer concentrations from 28
to 30 September attributed to southern Kalimantan. Be-
tween these aerosol events, a clean marine event from
26 until 28 September was characterized by high con-
centrations of Cl and low levels of elements associated
with anthropogenic sources. Back trajectories showed
that air masses traveled through the open, central SCS,
which suggests a good signal of sea spray was sampled.
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1271
As the ship route brought the Vasco near islands, local
crustal emissions covaried with sea spray aerosol, which
resulted in the crustal–marine mixed source during the
PMF analysis.
3. Six sources across the three size modes were resolved
by the PMF analysis: biomass burning, oil combustion,
soil dust, crustal–marine mixed source, sea spray, and
fly ash. A threshold Pearson Rcoefficient of 0.0 was
used to filter species included in the PMF analysis to im-
prove the interpretability of the PMF solution. Results
show that natural sources, the crustal–marine mixed
source and soil dust factors, were observed in only the
coarse and fine modes, while anthropogenic sources,
i.e., biomass burning, oil combustion, and fly ash, were
resolved purely in the fine and ultrafine modes. A strong
correspondence between key elements seen on the PMF
source profiles and groupings of these elements on the
correlation matrices adds confidence to the PMF solu-
tion. The biomass burning PMF factor showed the high-
est percent contributions to total elemental PM mass in
the fine and ultrafine modes: 32.4 % in the fine mode
and 45.9 % in the ultrafine mode. It is interesting to note
that the relative contribution of the oil combustion fac-
tor increased significantly towards finer modes, 7.3 %
in the fine mode but 26.1 % in the ultrafine mode, cor-
roborating its anthropogenic identification. In terms of
aerosol events, PMF source contributions were able to
capture the most events seen in the raw elemental con-
centrations. Differences in the temporal variations be-
tween PMF-resolved sources suggest these sources are
distinct. However, PMF did not differentiate between an
early ultrafine Si spike from a distinct, subsequent spike
in V, which demonstrates that PMF may merge events,
leading to a loss in resolution as observed in other stud-
ies (Van Pinxteren et al., 2016). This, however, can be
ameliorated with an in-depth, supervised analysis of the
data as has been done in this study.
4. As stated above, spikes in oil combustion tracers V and
Ni were observed on 18 September in the fine and ultra-
fine modes. HYSPLIT back trajectories suggest the ori-
gin of the air mass as Sandakan, an industrial area and
port city of Sabah known for its oil depots and shipping
activity located along the northeastern coast of Borneo.
The spike in oil combustion suggests that a small city
can cause drastic increases in tracer concentration de-
pending on air mass trajectories. The strong presence of
ultrafine mode Si from 18 to 19 September was also ob-
served, but the time series of Si is distinct from the time
series of V and Ni, suggestive of a source distinct from
oil combustion.
5. The 24 to 26 September event coincided with the ar-
rival of TC Nesat east of Luzon (northeast of the Vasco’s
location). Enhancements of multiple key tracers for
biomass burning, oil combustion, and soil dust were
observed, indicative of aerosols mixing within an air
mass during transport. Biomass burning tracers K, S,
Si, and Al show enhancements over a wider period (24–
26 September) than those of oil combustion tracers V
and Ni, which spiked at the end of the period. Further-
more, aerosol–convection interactions were observed as
sharp dips in the concentrations of biomass burning and
soil dust tracers around 25 September before recovery.
Interestingly, this dip was not observed for oil combus-
tion tracers V and Ni. This cold pool event was re-
ported in detail by Reid et al. (2015), and this study
further elaborated on its impact on PM of different ele-
mental compositions. This case demonstrates the effect
of short-term or high-frequency phenomena on aerosol
transport in the MC. HYSPLIT back trajectories show
that air masses begin to travel from the southwest MC
in response to TC Nesat’s inflow arm. Air masses during
the 24–26 September event pass through Brunei, a ship-
ping hub located along the northeastern coast of Borneo,
which explains the increase in oil combustion tracers V
and Ni. The coast was also observed to host a number
of active fire hotspots. Land breeze may lead to the en-
trainment of burning plumes into the traveling air mass,
which would explain the enrichment.
6. The 28–30 September aerosol event showed an enrich-
ment in K and S that coincided with a shift in back
trajectory origin to southern Kalimantan, which hosts
a high fire hotspot density. MC burning may be charac-
terized by an elevated K /S ratio and strong fine and ul-
trafine mode peaks in the mass distributions of S and K.
The 28–30 September event also coincided with the en-
hancement of soil dust elements in the coarse mode, in-
dicative of soil particle entrainment during burning ac-
tivity (Reid et al., 2015).
The study identified source locations of aerosol and char-
acterized the plumes during the Vasco 2011 cruise; however,
unanswered questions remain, such as the origin of the strong
ultrafine Si signal detected early in the cruise (18–19 Septem-
ber), which may be connected to a rapid local nucleation
event. The source location of the PMF-resolved fly ash fac-
tor also remains unidentified due to its complicated source
contribution time series and unclear elemental profile. Inves-
tigation into cloud nuclei (CN) properties during the cruise
may be done to further validate the intensity and timing of
plumes. In addition to the findings of this study on the ele-
mental PM, future research on other species collected dur-
ing the 2011 and 2012 Vasco campaigns such as trace gases
may complement and deepen our current understanding of
the aerosol environment in the SCS through additional de-
grees of freedom, specifically utilizing the lifetimes of trace
gases and inferring the potential for secondary aerosol for-
mation during transport.
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1272 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
Data availability. The Vasco ship data are available through cor-
respondence with Jeffrey S. Reid (jeffrey.reid@nrlmry.navy.mil).
MODIS AOD images were obtained from the NASA World-
view application: https://worldview.earthdata.nasa.gov/?v=74.
90262917418548,-13.851043766027335,166.53020142703977,
27.52453182940219&t=2011-09-17-T01:18:44Z&l=MODIS_
Terra_Aerosol,MODIS_Aqua_Aerosol,Reference_Labels(hidden)
,Reference_Features(hidden),Coastlines,VIIRS_SNPP_
CorrectedReflectance_TrueColor(hidden),MODIS_Aqua_
CorrectedReflectance_TrueColor(hidden),MODIS_Terra_
CorrectedReflectance_TrueColor (last access: 26 January 2020).
HYSPLIT data are accessible through the NOAA READY
website (https://www.ready.noaa.gov/hypub- bin/trajasrc.pl, last
access: 26 January 2020, NOAA Air Resources Laboratory,
2020). NAAPS aerosol reanalysis data can be accessed at
the US GODAE server: https://nrlgodae1.nrlmry.navy.mil/ftp/
outgoing/nrl/NAAPS-REANALYSIS/2011/201109/ (last access:
26 January 2020, Office of Naval Research, 2020).
Supplement. The supplement related to this article is available on-
line at: https://doi.org/10.5194/acp-20-1255-2020-supplement.
Author contributions. MRAH performed the analysis and prepared
the manuscript. MTC supervised the analysis, especially for the
PMF section. MOLC supervised the analysis and provided input
for the manuscript. JSR collected the data on board the Vasco, su-
pervised the analysis, and provided input for the manuscript. PX
provided the NAAPS smoke model outputs for Fig. 2 and provided
input for the manuscript. JBS, NDL, and SNYU collected the data
on board the Vasco. SC and YJZ performed the XRF analysis on the
data.
Competing interests. The authors declare that they have no conflict
of interest.
Acknowledgements. We acknowledge the use of imagery from the
NASA Worldview application (https://worldview.earthdata.nasa.
gov/, last access: 28 April 2018), part of the NASA Earth Ob-
serving System Data and Information System (EOSDIS). The au-
thors gratefully acknowledge the NOAA Air Resources Labora-
tory (ARL) for the provision of the HYSPLIT transport and disper-
sion model and/or READY website (https://www.ready.noaa.gov/
hypub-bin/trajasrc.pl, last access: 26 January 2020) used in this
publication.
Review statement. This paper was edited by Tuukka Petäjä and re-
viewed by two anonymous referees.
References
Alam, K., Mukhtar, A., Shahid, I., Blaschke, T., Majid, H., Rah-
man, S., Khan, R., and Rahman, N.: Source apportionment
and characterization of particulate matter (PM10) in urban en-
vironment of Lahore, Aerosol Air Qual. Res., 14, 1851–1861,
https://doi.org/10.4209/aaqr.2014.01.0005, 2014.
Artaxo, P. and Maenhaut, W.: Aerosol Characteristics and Sources
for the Amazon Basin During the Wet Season, J. Geophys. Res.,
95, 16971–16985, 1990.
Artaxo, P., Fernandas, E. T., Martins, J. V., Yamasoe, M.
A., Maenhaut, W., Longo, K. M., Castanho, A., and
Hobbs, P. V.: Large-scale aerosol source apportionment in
Amazonia, J. Geophys. Res.-Atmos., 103, 31837–31847,
https://doi.org/10.1029/98jd02346, 1998.
Atwood, S. A., Reid, J. S., Kreidenweis, S. M., Cliff, S. S., Zhao,
Y., Lin, N.-H., Tsay, S.-C., Chu, Y.-C., and Westphal, D. L.:
Size resolved measurements of springtime aerosol particles over
the northern South China Sea, Atmos. Environ., 78, 134–143,
https://doi.org/10.1016/j.atmosenv.2012.11.024, 2013a.
Atwood, S. A., Salinas, S. V., Chew, B. N., Reid, J. S., Bala-
subramanian, R., Yu, L. E., and Kreidenweis, S. M.: Analy-
sis of source regions for smoke events in Singapore for the
2009 El Nino burning season, Atmos. Environ., 78, 219–230,
https://doi.org/10.1016/j.atmosenv.2013.04.047, 2013b.
Atwood, S. A., Reid, J. S., Kreidenweis, S. M., Blake, D. R., Jon-
sson, H. H., Lagrosas, N. D., Xian, P., Reid, E. A., Sessions,
W. R., and Simpas, J. B.: Size-resolved aerosol and cloud con-
densation nuclei (CCN) properties in the remote marine South
China Sea – Part 1: Observations and source classification, At-
mos. Chem. Phys., 17, 1105–1123, https://doi.org/10.5194/acp-
17-1105-2017, 2017.
Balasubramanian, R., Qian, W.-B., Decesari, S., Facchini,
M. C., and Fuzzi, S.: Comprehensive characterization of
PM2.5aerosols in Singapore, J. Geophys. Res., 108, 4523,
https://doi.org/10.1029/2002jd002517, 2003.
Becagli, S., Sferlazzo, D. M., Pace, G., di Sarra, A., Bommarito,
C., Calzolai, G., Ghedini, C., Lucarelli, F., Meloni, D., Mon-
teleone, F., Severi, M., Traversi, R., and Udisti, R.: Evidence
for heavy fuel oil combustion aerosols from chemical analyses
at the island of Lampedusa: a possible large role of ships emis-
sions in the Mediterranean, Atmos. Chem. Phys., 12, 3479–3492,
https://doi.org/10.5194/acp-12-3479-2012, 2012.
Boucher, O., Randall, D. Artaxo, P., Bretherton, C., Feingold, G.,
Forster, P., Kerminen, V.-M., Kondo, Y., Liao, H., Lohmann, U.,
Rasch, P., Satheesh, S. K., Sherwood, S., Stevens, B., and Zhang,
X. Y.: Clouds and Aerosols, in: Climate Change 2013: The Phys-
ical Science Basis, Contribution of Working Group I to the Fifth
Assessment Report of the Intergovernmental Panel on Climate
Change, edited by: Stocker, T. F., Qin, D., Plattner, G.-K., Tig-
nor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex,
V., and Midgley, P. M., Cambridge University Press, Cambridge,
UK, New York, NY, USA, 2013.
Cahill, T. A., Barnes, D. E., Lawton, J. A., Miller, R., Spada, N.,
C, R. D. W. and Kimbrough, S.: Transition metals in coarse,
fine, very fine and ultra-fine particles from an interstate high-
way transect near Detroit, Adv. Stud. Behav., 37, 371–401,
https://doi.org/10.1016/S0065-3454(07)37008-3, 2007.
Cahill, T. A., Barnes, D. E., Lawton, J. A., Miller, R., Spada,
N., Willis, R. D., and Kimbrough, S.: Transition metals in
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1273
coarse, fine, very fine and ultra-fine particles from an interstate
highway transect near Detroit, Atmos. Environ., 145, 158–175,
https://doi.org/10.1016/j.atmosenv.2016.09.023, 2016.
Chan, Y. C., Christensen, E., Golding, G., King, G., Gore, W., Co-
hen, D. D., Stelcer, E., Simpson, R., Denison, L., and Wong, N.:
Source Apportionment of Ambient Volatile Organic Compounds
in Major Cities in Australia By Positive Matrix Factorisation,
Clean Air and Environmental Quality, 42, 22–29, 2008.
Chang, C. P., Zhuo, W., John, M., and Ching-Hwang, L.:
Annual Cycle of Southeast Asia – Maritime Continent
Rainfall and the Asymmetric, J. Climate, 18, 287–301,
https://doi.org/10.1175/JCLI-3257.1, 2005.
Chow, J. C., Lowenthal, D. H., Chen, L.-W. A., Wang,
X., and Watson, J. G.: Mass reconstruction methods for
PM2.5: a review, Air Qual. Atmos. Health, 8, 243–263,
https://doi.org/10.1007/s11869-015-0338-3, 2015.
Cohen, D. D., Crawford, J., Stelcer, E., and Bac, V. T.:
Long range transport of fine particle windblown soils
and coal fired power station emissions into Hanoi be-
tween 2001 to 2008, Atmos. Environ., 44, 3761–3769,
https://doi.org/10.1016/j.atmosenv.2010.06.047, 2010a.
Cohen, D. D., Crawford, J., Stelcer, E., and Bac, V. T.: Charac-
terisation and source apportionment of fine particulate sources
at Hanoi from 2001 to 2008, Atmos. Environ., 44, 320–328,
https://doi.org/10.1016/j.atmosenv.2009.10.037, 2010b.
Davison, R. L., Natusch, D. F. S., and Wallace, J. R.: Trace
Elements in Fly Ash, Environ. Sci. Technol., 8, 1107–1113,
https://doi.org/10.1021/es60098a003, 1974.
Deonarine, A., Kolker, A., and Doughten, M.: Trace Elements in
Coal Ash, U.S. Geological Survey Fact Sheet 2015–3037, 6 pp.,
https://doi.org/10.3133/fs20153037, 2015.
Draxler, R. R. and Hess, G. D.: An overview of the HYSPLIT_4
modelling system for trajectories, Aust. Meteorol. Mag., 47,
295–308, 1998.
Draxler, R. R.: HYSPLIT4 user’s guide. NOAA Tech. Memo. ERL
ARL-230, NOAA Air Resources Laboratory, Silver Spring, MD,
1999.
Engling, G., He, J., Betha, R., and Balasubramanian, R.: As-
sessing the regional impact of indonesian biomass burning
emissions based on organic molecular tracers and chemical
mass balance modeling, Atmos. Chem. Phys., 14, 8043–8054,
https://doi.org/10.5194/acp-14-8043-2014, 2014.
Farao, C., Canepari, S., Perrino, C., and Harrison, R. M.:
Sources of PM in an industrial area: Comparison between
receptor model results and semiempirical calculations of
source contributions, Aerosol Air Qual. Res., 14, 1558–1572,
https://doi.org/10.4209/aaqr.2013.08.0281, 2014.
Field, R. D. and Shen, S. S. P.: Predictability of car-
bon emissions from biomass burning in Indonesia from
1997 to 2006, J. Geophys. Res.-Biogeo., 113, 1–17,
https://doi.org/10.1029/2008JG000694, 2008.
Fujii, Y., Tohno, S., Amil, N., Latif, M. T., Oda, M., Matsumoto, J.,
and Mizohata, A.: Annual variations of carbonaceous PM2.5in
Malaysia: influence by Indonesian peatland fires, Atmos. Chem.
Phys., 15, 13319–13329, https://doi.org/10.5194/acp-15-13319-
2015, 2015.
Ge, C., Wang, J., and Reid, J. S.: Mesoscale modeling of
smoke transport over the Southeast Asian Maritime Conti-
nent: coupling of smoke direct radiative effect below and
above the low-level clouds, Atmos. Chem. Phys., 14, 159–174,
https://doi.org/10.5194/acp-14-159-2014, 2014.
Gugamsetty, B., Wei, H., Liu, C. N., Awasthi, A., Tsai, C. J., Roam,
G. D., Wu, Y. C., and Chen, C. F.: Source Characterization
and Apportionment of PM10, PM2.5and PM0.1by Using Pos-
itive Matrix Factorization, Aerosol Air Qual. Res., 12, 476–491,
https://doi.org/10.4209/aaqr.2012.04.0084, 2012.
Han, J. S., Moon, K. J., Lee, S. J., Kim, Y. J., Ryu, S. Y., Cliff,
S. S., and Yi, S. M.: Size-resolved source apportionment of am-
bient particles by positive matrix factorization at Gosan back-
ground site in East Asia, Atmos. Chem. Phys., 6, 211–223,
https://doi.org/10.5194/acp-6-211-2006, 2006.
Hedberg, E., Gidhagen, L., and Johansson, C.: Source contribu-
tions to PM10 and arsenic concentrations in Central Chile us-
ing positive matrix factorization, Atmos. Environ., 39, 549–561,
https://doi.org/10.1016/j.atmosenv.2004.11.001, 2005.
Hyer, E. J., Reid, J. S., and Zhang, J.: An over-land aerosol opti-
cal depth data set for data assimilation by filtering, correction,
and aggregation of MODIS Collection 5 optical depth retrievals,
Atmos. Meas. Tech., 4, 379–408, https://doi.org/10.5194/amt-4-
379-2011, 2011.
Karanasiou, A. A., Siskos, P. A., and Eleftheriadis, K.:
Assessment of source apportionment by Positive Ma-
trix Factorization analysis on fine and coarse urban
aerosol size fractions, Atmos. Environ., 43, 3385–3395,
https://doi.org/10.1016/j.atmosenv.2009.03.051, 2009.
Landis, M. S., Pancras, J. P., Graney, J. R., White, E. M., Edger-
ton, E. S., Legge, A., and Percy, K. E.: Source apportion-
ment of ambient fine and coarse particulate matter at the
Fort McKay community site, in the Athabasca Oil Sands Re-
gion, Alberta, Canada, Sci. Total Environ., 584–585, 105–117,
https://doi.org/10.1016/j.scitotenv.2017.01.110, 2017.
Lee, E., Chan, C. K., and Paatero, P.: Application of posi-
tive matrix factorization in source apportionment of particu-
late pollutants in Hong Kong, Atmos. Environ., 33, 3201–3212,
https://doi.org/10.1016/S1352-2310(99)00113-2, 1999.
Lee, S.-S., Feingold, G., and Chuang, P. Y.: Effect of Aerosol
on Cloud–Environment Interactions in Trade Cumulus, J.
Atmos. Sci., 69, 3607–3632, https://doi.org/10.1175/jas-d-12-
026.1, 2012.
Lestari, P. and Mauliadi, Y. D.: Source apportionment
of particulate matter at urban mixed site in Indone-
sia using PMF, Atmos. Environ., 43, 1760–1770,
https://doi.org/10.1016/j.atmosenv.2008.12.044, 2009.
Liao, H.-T., Chang, J.-C., Tsai, T.-T., Tsai, S.-W., Chou, C. C.-K.,
and Wu, C.-F.: Vertical distribution of source apportioned PM2.5
using particulate-bound elements and polycyclic aromatic hydro-
carbons in an urban area, J. Expo Sci. Environ. Epidemiol., 1–11,
https://doi.org/10.1038/s41370-019-0153-2, 2019.
Lin, C.-Y., Wang, Z., Chen, W.-N., Chang, S.-Y., Chou, C.
C. K., Sugimoto, N., and Zhao, X.: Long-range transport
of Asian dust and air pollutants to Taiwan: observed evi-
dence and model simulation, Atmos. Chem. Phys., 7, 423–434,
https://doi.org/10.5194/acp-7-423-2007, 2007.
Lin, G., Penner, J. E., Flanner, M. G., Sillman, S., Xu,
L. and Zhou, C.: Radiative forcing of organic aerosol
in the atmosphere and on snow: Effects of SOA and
brown carbon, J. Geophys. Res.-Atmos., 119, 7453–7476,
https://doi.org/10.1002/2013JD021186, 2014.
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1274 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
Lin, Y.-C., Tsai, C.-J., Wu, Y.-C., Zhang, R., Chi, K.-H., Huang,
Y.-T., Lin, S.-H., and Hsu, S.-C.: Characteristics of trace met-
als in traffic-derived particles in Hsuehshan Tunnel, Taiwan: size
distribution, potential source, and fingerprinting metal ratio, At-
mos. Chem. Phys., 15, 4117–4130, https://doi.org/10.5194/acp-
15-4117-2015, 2015.
Loo, Y. Y., Billa, L., and Singh, A.: Effect of climate change on
seasonal monsoon in Asia and its impact on the variability of
monsoon rainfall in Southeast Asia, Geosci. Front., 6, 817–823,
https://doi.org/10.1016/j.gsf.2014.02.009, 2015.
Lynch, P., Reid, J. S., Westphal, D. L., Zhang, J., Hogan, T. F., Hyer,
E. J., Curtis, C. A., Hegg, D. A., Shi, Y., Campbell, J. R., Rubin,
J. I., Sessions, W. R., Turk, F. J., and Walker, A. L.: An 11-year
global gridded aerosol optical thickness reanalysis (v1.0) for at-
mospheric and climate sciences, Geosci. Model Dev., 9, 1489–
1522, https://doi.org/10.5194/gmd-9-1489-2016, 2016.
Ma, L., Dadashazar, H., Braun, R. A., MacDonald, A. B.,
Aghdam, M. A., Maudlin, L. C., and Sorooshian, A.: Size-
resolved characteristics of water-soluble particulate elements
in a coastal area: Source identification, influence of wild-
fires, and diurnal variability, Atmos. Environ., 206, 72–84,
https://doi.org/10.1016/j.atmosenv.2019.02.045, 2019.
Malm, W. C. and Hand, J. L.: An examination of the
physical and optical properties of aerosols collected in
the IMPROVE program, Atmos. Environ., 41, 3407–3427,
https://doi.org/10.1016/j.atmosenv.2006.12.012, 2007.
Mazzei, F., D’Alessandro, A., Lucarelli, F., Nava, S., Prati, P.,
Valli, G., and Vecchi, R.: Characterization of particulate matter
sources in an urban environment, Sci. Total Environ., 401, 81–89,
https://doi.org/10.1016/j.scitotenv.2008.03.008, 2008.
Markowski, G. R. and Filby, R.: Trace Element Concentration
as a Function of Particle Size in Fly Ash from a Pulver-
ized Coal Utility Boiler, Environ. Sci. Technol., 19, 796–804,
https://doi.org/10.1021/es00139a005, 1985.
Nakajima, T., Yoon, S. C., Ramanathan, V., Shi, G. Y., Takemura,
T., Higurashi, A., Takamura, T., Aoki, K., Sohn, B. J., Kim,
S. W., Tsuruta, H., Sugimoto, N., Shimizu, A., Tanimoto, H.,
Sawa, Y., Lin, N. H., Lee, C. Te, Goto, D., and Schutgens,
N.: Overview of the Atmospheric Brown Cloud East Asian Re-
gional Experiment 2005 and a study of the aerosol direct radia-
tive forcing in east Asia, J. Geophys. Res.-Atmos., 112, 1–23,
https://doi.org/10.1029/2007JD009009, 2007.
Nigam, A., Welch, W., Wayne Miller, J., and Cocher III, D. R.:
Effect of fuel sulphur content and control technology on PM
emission from ship’s auxiliary engine. Proceeding international
aerosol conference, St. Paul, USA, 10–15 September 2006,
1531–1532, 2006.
NOAA Air Resources Laboratory: Air Resources Laboratory
– HYSPLIT – Hybrid Single Particle Lagrangian Inte-
grated Trajectory model, Real-time Environmental Applications
and Display sYstem, available at: https://www.ready.noaa.gov/
HYSPLIT.php, last access: 26 January 2020.
Office of Naval Research: Index of /ftp/outgoing/nrl/NAAPS-
REANALYSIS/2011/201109, Global Ocean Data Assimilation
Experiment, available at: https://nrlgodae1.nrlmry.navy.mil/ftp/
outgoing/nrl/NAAPS-REANALYSIS/2011/201109/, last access:
26 January 2020.
O’Sullivan, D., Murray, B. J., Malkin, T. L., Whale, T. F., Umo,
N. S., Atkinson, J. D., Price, H. C., Baustian, K. J., Browse,
J., and Webb, M. E.: Ice nucleation by fertile soil dusts: rel-
ative importance of mineral and biogenic components, Atmos.
Chem. Phys., 14, 1853–1867, https://doi.org/10.5194/acp-14-
1853-2014, 2014.
Paatero, P. and Tapper, U.: Positive matrix factorization: A
non-negative factor model with optimal utilization of er-
ror estimates of data values, Environmetrics, 5, 111–126,
https://doi.org/10.1002/env.3170050203, 1994.
Paatero, P. and Hopke, P. K.: Discarding or downweighting high-
noise variables in factor analytic models, Anal. Chim. Acta,
490, 277–289, https://doi.org/10.1016/S0003-2670(02)01643-4,
2003.
Paterson, K. G., Sagady, J. L., Hooper, D. L., Bertman, S. B., Car-
roll, M. A., and Shepson, P. B.: Analysis of air quality data using
positive matrix factorization, Environ. Sci. Technol., 33, 635–
641, https://doi.org/10.1021/es980605j, 1999.
Polissar, A. V., Hopke, P. K., Paatero, P., Malm, W. C., and Sisler,
J. F.: Atmospheric aerosol over Alaska 2. Elemental composi-
tion and sources, J. Geophys. Res.-Atmos., 103, 19045–19057,
https://doi.org/10.1029/98JD01212, 1998.
Prospero, J. M.: Long-range transport of mineral dust in the global
atmosphere: Impact of African dust on the environment of the
southeastern United States, P. Natl. Acad. Sci. USA, 96, 3396–
3403, https://doi.org/10.1073/pnas.96.7.3396, 1999.
Raabe, O. G., Braaten, D. A., Axelbaum, R. L., Teague, S. V., and
Cahill, T. A.: Calibration studies of the drum impactor, J. Aerosol
Sci., 19, 183–195, https://doi.org/10.1016/0021-8502(88)90222-
4, 1988.
Reid, J. S., Flocchini, R. G., Cahill, T. A., Ruth, R. S., and Sal-
gado, D. P.: Local meteorological, transport, and source aerosol
characteristics of late autumn Owens Lake (dry) dust storms,
Atmos. Environ., 28, 1699–1706, https://doi.org/10.1016/1352-
2310(94)90315-8, 1994.
Reid, J. S., Hyer, E. J., Prins, E. M., Westphal, D. L., Zhang,
J., Wang, J., Christopher, S. A., Curtis, C. A., Schmidt,
C. C., Eleuterio, D. P., Richardson, K. A., and Hoffman,
J. P.: Global monitoring and forecasting of biomass-burning
smoke: Description of and lessons from the fire Locating and
Modeling of Burning Emissions (FLAMBE) program, IEEE
J. Sel. Top. Appl. Earth Obs. Remote Sens., 2, 144–162,
https://doi.org/10.1109/JSTARS.2009.2027443, 2009.
Reid, J. S., Xian, P., Hyer, E. J., Flatau, M. K., Ramirez, E. M.,
Turk, F. J., Sampson, C. R., Zhang, C., Fukada, E. M., and Mal-
oney, E. D.: Multi-scale meteorological conceptual analysis of
observed active fire hotspot activity and smoke optical depth in
the Maritime Continent, Atmos. Chem. Phys., 12, 2117–2147,
https://doi.org/10.5194/acp-12-2117-2012, 2012.
Reid, J., Hyer, E., Johnson, R., Holben, B. N., Yokelson, R.,
Zhang, J., Campbell, J., Christopher, S., DiGirolamo, L., Giglio,
L., Holz, R., Kearney, C., Miettinen, J., Reid, E., Turk, F. J.,
Wang, J., Xian, P., Zhao, G., Balasubramanian, R., Chew, B.
N., Janjai, S., Lagrosas, N., Lestari, P., Lin, N.-H., Mahmud,
M., Nguyen, A. X., Norris, B., Oanth, N. T. K., Oo, M., Sali-
nas, S. V., Welton, E. J., and Liew, S. C.: Observing and un-
derstanding the Southeast Asian aerosol system by remote sens-
ing: An initial review and analysis for the Seven Southeast
Asian Studies (7SEAS) program, Atmos. Res., 122, 403–468,
https://doi.org/10.1016/j.atmosres.2012.06.005, 2013.
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/

M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS 1275
Reid, J. S., Lagrosas, N. D., Jonsson, H. H., Reid, E. A., Sessions,
W. R., Simpas, J. B., Uy, S. N., Boyd, T. J., Atwood, S. A., Blake,
D. R., Campbell, J. R., Cliff, S. S., Holben, B. N., Holz, R. E.,
Hyer, E. J., Lynch, P., Meinardi, S., Posselt, D. J., Richardson,
K. A., Salinas, S. V., Smirnov, A., Wang, Q., Yu, L., and Zhang,
J.: Observations of the temporal variability in aerosol properties
and their relationships to meteorology in the summer monsoonal
South China Sea/East Sea: the scale-dependent role of mon-
soonal flows, the Madden–Julian Oscillation, tropical cyclones,
squall lines and cold pools, Atmos. Chem. Phys., 15, 1745–1768,
https://doi.org/10.5194/acp-15-1745-2015, 2015.
Reid, J. S., Xian, P., Holben, B. N., Hyer, E. J., Reid, E. A., Sali-
nas, S. V., Zhang, J., Campbell, J. R., Chew, B. N., Holz, R. E.,
Kuciauskas, A. P., Lagrosas, N., Posselt, D. J., Sampson, C. R.,
Walker, A. L., Welton, E. J., and Zhang, C.: Aerosol meteorology
of the Maritime Continent for the 2012 7SEAS southwest mon-
soon intensive study – Part 1: regional-scale phenomena, Atmos.
Chem. Phys., 16, 14041–14056, https://doi.org/10.5194/acp-16-
14041-2016, 2016.
Ross, A. D., Holz, R. E., Quinn, G., Reid, J. S., Xian, P., Turk,
F. J., and Posselt, D. J.: Exploring the first aerosol indirect ef-
fect over Southeast Asia using a 10-year collocated MODIS,
CALIOP, and model dataset, Atmos. Chem. Phys., 18, 12747–
12764, https://doi.org/10.5194/acp-18-12747-2018, 2018.
Rolph, G., Stein, A., and Stunder, B.: Real-time Environmental Ap-
plications and Display sYstem: READY, Environ. Model. Softw.,
95, 210–228, https://doi.org/10.1016/j.envsoft.2017.06, 2017.
Santoso, M., Lestiani, D. D., Mukhtar, R., Hamonangan,
E., Syafrul, H., Markwitz, A., and Hopke, P. K.: Pre-
liminary study of the sources of ambient air pollution
in Serpong, Indonesia, Atmos. Pollut. Res., 2, 190–196,
https://doi.org/10.5094/apr.2011.024, 2011.
Schlosser, J. S., Braun, R. A., Bradley, T., Dadashazar, H., Mac-
Donald, A. B., Aldhaif, A. A., Aghdam, M. A., Mardi, A.
H., Xian, P., and Sorooshian, A.: Analysis of aerosol composi-
tion data for western United States wildfires between 2005 and
2015: Dust emissions, chloride depletion, and most enhanced
aerosol constituents, J. Geophys. Res.-Atmos., 122, 8951–8966,
https://doi.org/10.1002/2017JD026547, 2017.
Seneviratne, M. C. S., Waduge, V. A., Hadagiripathira, L., San-
jeewani, S., Attanayake, T., Jayaratne, N., and Hopke, P. K.:
Characterization and source apportionment of particulate pollu-
tion in Colombo, Sri Lanka, Atmos. Pollut. Res., 2, 207–212,
https://doi.org/10.5094/apr.2011.026, 2011.
Shi, Y., Zhang, J., Reid, J. S., Liu, B., and Hyer, E. J.: Critical eval-
uation of cloud contamination in the MISR aerosol products us-
ing MODIS cloud mask products, Atmos. Meas. Tech., 7, 1791–
1801, https://doi.org/10.5194/amt-7-1791-2014, 2014.
Song, Y., Xie, S., Zhang, Y., Zeng, L., Salmon, L. G., and
Zheng, M.: Source apportionment of PM2.5in Beijing us-
ing principal component analysis/absolute principal compo-
nent scores and UNMIX, Sci. Total Environ., 372, 278–286,
https://doi.org/10.1016/j.scitotenv.2006.08.041, 2006.
Sorooshian, A., Feingold, G., Lebsock, M. D., Jiang, H., and
Stephens, G. L.: On the precipitation susceptibility of clouds
to aerosol perturbations, Geophys. Res. Lett., 36, L13803,
https://doi.org/10.1029/2009GL038993, 2009.
Stein, A. F., Draxler, R. R, Rolph, G. D., Stunder, B. J. B., Cohen,
M. D., and Ngan, F.: NOAA’s HYSPLIT atmospheric transport
and dispersion modeling system, B. Am. Meteor. Soc., 96, 2059–
2077, https://doi.org/10.1175/BAMS-D-14-00110.1, 2015.
Tian, S. L., Pan, Y. P., and Wang, Y. S.: Size-resolved source
apportionment of particulate matter in urban Beijing during
haze and non-haze episodes, Atmos. Chem. Phys., 16, 1–19,
https://doi.org/10.5194/acp-16-1-2016, 2016.
Twohy, C. H., Kreidenweis, S. M., Eidhammer, T., Browell, E. V.,
Heymsfield, A. J., Bansemer, A. R., Anderson, B. E., Chen, G.,
Ismail, S., DeMott, P. J., and Van Den Heever, S. C.: Saharan dust
particles nucleate droplets in eastern Atlantic clouds, Geophys.
Res. Lett., 36, L01807, https://doi.org/10.1029/2008GL035846,
2009.
Van Pinxteren, D., Fomba, K. W., Spindler, G., Müller, K., Poulain,
L., Iinuma, Y., Löschau, G., Hausmann, A., and Herrmann,
H.: Regional air quality in Leipzig, Germany: Detailed source
apportionment of size-resolved aerosol particles and compar-
ison with the year 2000, Faraday Discuss., 189, 291–315,
https://doi.org/10.1039/c5fd00228a, 2016.
Venecek, M. A., Zhao, Y., Mojica, J., McDade, C. E., Green,
P. G., Kleeman, M. J., and Wexler, A. S.: Characteri-
zation of the 8-stage Rotating Drum Impactor under low
concentration conditions, J. Aerosol Science, 100, 140–154,
https://doi.org/10.1016/j.jaerosci.2016.07.007, 2016.
Viana, M., Kuhlbusch, T. A. J., Querol, X., Alastuey, A., Harri-
son, R. M., Hopke, P. K., Winiwarter, W., Vallius, M., Szidat, S.,
Prévôt, A. S. H., Hueglin, C., Bloemen, H., Wåhlin, P., Vecchi,
R., Miranda, A. I., Kasper-Giebl, A., Maenhaut, W., and Hitzen-
berger, R.: Source apportionment of particulate matter in Europe:
A review of methods and results, J. Aerosol Sci., 39, 827–849,
https://doi.org/10.1016/j.jaerosci.2008.05.007, 2008.
Wang, S.-H., Tsay, S.-C., Lin, N.-H., Hsu, N. C., Bell, S. W., Li,
C., Ji, Q., Jeong, M.-J., Hansell, R. A., Welton, E. J., Holben, B.
N., Sheu, G.-R., Chu, Y.-C., Chang, S.-C., Liu, J.-J., and Chiang,
W.-L.: First detailed observations of long-range transported dust
over the northern South China Sea, Atmos. Environ., 45, 4804–
4808, https://doi.org/10.1016/j.atmosenv.2011.04.077, 2011.
Wang, Z., Sorooshian, A., Prabhakar, G., Coggon, M. M., and Jon-
sson, H. H.: Impact of emissions from shipping, land, and the
ocean on stratocumulus cloud water elemental composition dur-
ing the 2011 E-PEACE field campaign,
Wimolwattanapun, W., Hopke, P. K. and Pongkiatkul, P.: Source
apportionment and potential source locations of PM2.5and
PM2.5−10 at residential sites in metropolitan Bangkok, Atmos.
Pollut. Res., 2, 172–181, https://doi.org/10.5094/apr.2011.022,
2011.
Witt, M., Baker, A. R., and Jickells, T. D.: Atmospheric trace
metals over the Atlantic and South Indian Oceans: Investiga-
tion of metal concentrations and lead isotope ratios in coastal
and remote marine aerosols, Atmos. Environ., 40, 5435–5451,
https://doi.org/10.1016/j.atmosenv.2006.04.041, 2006.
Xian, P., Reid, J. S., Atwood, S. A., Johnson, R. S., Hyer, E. J.,
Westphal, D. L., and Sessions, W.: Smoke aerosol transport pat-
terns over the Maritime Continent, Atmos. Res., 122, 469–485,
https://doi.org/10.1016/j.atmosres.2012.05.006, 2013.
Xie, R. K., Seip, H. M., Liu, L., and Zhang, D. S.: Characterization
of individual airborne particles in Taiyuan City, China, Air Qual.
Atmos. Heal., 2, 123–131, https://doi.org/10.1007/s11869-009-
0039-x, 2009.
www.atmos-chem-phys.net/20/1255/2020/ Atmos. Chem. Phys., 20, 1255–1276, 2020

1276 M. R. A. Hilario et al.: Investigating size-segregated sources of elemental composition of PM in the SCS
Yusuf, A. A. and Francisco, H.: Climate change vulnerability map-
ping for Southeast Asia, Economy and Environment Program
for Southeast Asia (EEPSEA) report, available at: http://www.
eepsea.net/ (last access: July 2014), 32 pp., 2009.
Zhang, J. and Reid, J. S.: MODIS Aerosol Product Analysis
for Data Assimilation: Assessment of Level 2 Aerosol Opti-
cal Thickness Retrievals, J. Geophys. Res.-Atmos., 111, 22207,
https://doi.org/10.1029/2005JD006898, 2006.
Zheng, C., Zhao, C., Zhu, Y., Wang, Y., Shi, X., Wu, X., Chen,
T., Wu, F., and Qiu, Y.: Analysis of influential factors for the
relationship between PM2.5and AOD in Beijing, Atmos. Chem.
Phys., 17, 13473–13489, https://doi.org/10.5194/acp-17-13473-
2017, 2017.
Atmos. Chem. Phys., 20, 1255–1276, 2020 www.atmos-chem-phys.net/20/1255/2020/