Introduction
Organic compounds comprise up to 90 % of the mass concentration of
submicron organic particulate matter (PM) over tropical forests (Kanakidou et
al., 2005). Submicron PM has adverse effects on human health (Nel, 2005; Pope III and Dockery, 2006)
and influences air quality and climate by scattering
radiation and acting as cloud condensation nuclei (Ramanathan et al., 2001;
Kaufman et al., 2002). A significant fraction of the submicron organic
material originates from secondary processes, mainly by the atmospheric
oxidation of volatile organic compounds (VOCs) emitted as part of natural and
human activities (Zhang et al., 2007; Hallquist et al., 2009; Jimenez et al.,
2009). The particle life cycle over Amazonia is in particular strongly
influenced by secondary processes that produce organic PM (Martin et al.,
2010a; Pöschl et al., 2010). Biogenic emissions from tropical forests are
high, and environmental conditions favor photooxidation reactions. The
reactive chemistry and the relative importance of pathways leading to PM
production can be strongly guided by regulating species, such as sulfate and
nitric oxide (NO) (Surratt et al., 2007a; Worton et al., 2013; Liu et al., 2016a).
The concentrations of these species depend on their background occurrence,
pollution sources, and the relative mix of background and polluted air
masses.
Schematic diagram for the production of PM derived from isoprene
epoxydiols (IEPOX) produced during the
photooxidation of isoprene. Organic peroxy radicals (ISOPOO), produced by OH
attack and O2 addition to isoprene, are scavenged along NO or HO2
pathways. By the HO2 pathway, organic hydroperoxides (ISOPOOH) are
first-generation products that react with additional OH to produce IEPOX. The IEPOX species undergo reactive uptake into
particles, ultimately producing IEPOX-derived particulate matter. Arrow
thickness qualitatively illustrates the relative importance (i.e., mass
flux) of a reaction channel under background conditions. Gray and green
background colors indicate species in the gas and particle phases,
respectively. The light-green disk represents the total organic PM. Within
that disk, the contribution by isoprene-derived PM, including compounds
produced both IEPOX and non-IEPOX pathways, is represented by the dark-green
oval. Inside that oval, the contribution by IEPOX-derived PM is represented
by the yellow oval region. The color gradient between brown and dark green
illustrates the chemical modification of the IEPOX-derived PM over time. The
large dashed black arrow represents the analytical methods that use
different types of molecular and statistical tracers (listed in the boxes)
to quantify the IEPOX-derived PM mass concentrations. For simplicity, the
figure omits the many routes leading to the production of glyoxal (Fu et
al., 2008), possible ISOPOO isomerization when NO and HO2
concentrations are sufficiently low (Crounse et al., 2011; Liu et al.,
2016a), second-generation production of peroxymethacrylic nitric anhydride
(Lin et al., 2013; Nguyen et al., 2015), and particle water and other
inorganic components. 3-methyltetrahydrofuran-3,4-diols are abbreviated as
3-MeTHF-3,4-diols. Other abbreviations are provided in the main text.
Over tropical forests such as Amazonia, the atmospheric chemistry of isoprene
produces a substantial fraction of the submicron organic PM (Chen et al.,
2009, 2015; Robinson et al., 2011; Isaacman-VanWertz et al.,
2016). Isoprene (2-methyl-1,3-butadiene, C5H8) is the non-methane
VOC most abundantly emitted by tropical forests (Guenther et al., 2012), and
isoprene epoxydiols (IEPOX) have been identified as important intermediates
in the production of PM from isoprene (Paulot et al., 2009; Surratt et al.,
2010; Lin et al., 2012). A chemical sequence for the production of
IEPOX-derived PM from the photooxidation of isoprene in the atmosphere is
represented in Fig. 1. The sequence is initiated when isoprene peroxy
radicals (ISOPOO) are produced in the gas phase by reactions between isoprene
and photochemically generated hydroxyl radicals (OH). The reactive fate of
the ISOPOO radicals can differ under background compared to polluted
conditions (Surratt et al., 2010; Crounse et al., 2011; Worton et al., 2013).
Under background conditions, meaning that HO2 pathways are favorable in
the absence of extensive NO pollution (Wennberg, 2013; Liu et al., 2016a),
the ISOPOO radicals continue in large part through the series of species
highlighted in yellow in Fig. 1. Through HOx-facilitated reaction steps,
the ISOPOO radicals produce hydroperoxides (ISOPOOH) as major
first-generation products and subsequently IEPOX as
major second-generation products (Carlton et al., 2009; Paulot et al., 2009;
Liu et al., 2013, 2016a; St. Clair et al., 2015). Some of the
produced IEPOX undergoes reactive uptake to particles, as facilitated by
hydronium ions at the surface (Surratt et al., 2007a; Lin et al., 2012;
Gaston et al., 2014; Kuwata et al., 2015; Lewandowski et al., 2015). This
chemical sequence can contribute a significant fraction of submicron PM mass
concentration over tropical forests (Claeys et al., 2004; Hu et al., 2015).
Laboratory studies indicate that about half of the PM produced by isoprene
photooxidation under HO2-dominant conditions in the presence of acidic
sulfate particles is associated with IEPOX production and uptake (Liu et al.,
2015). Interaction of IEPOX with cloud waters warrants investigation (Lim et
al., 2005; Ervens et al., 2011; Budisulistiorini et al., 2015; Chen et al.,
2015). In addition to IEPOX pathways, laboratory studies suggest that
multifunctional hydroperoxides produced in the gas phase can contribute to
isoprene-derived PM production (Krechmer et al., 2015; Liu et al., 2016b;
Riva et al., 2016b).
After reactive uptake of IEPOX, particle-phase reactions can produce several
different families of species. These species are collectively labeled
“IEPOX-derived PM” and represent a subset of the ambient organic PM, as
labeled in Fig. 1. The presence of 2-methyltetrols, C5-alkene triols,
3-methyltetrahydrofuran-3,4-diols, organosulfates, and related oligomers in
ambient PM is an indicator of PM production by IEPOX uptake under atmospheric
conditions (Claeys et al., 2004; Surratt et al., 2006, 2007b, 2010; Robinson et al., 2011; Lin et al., 2012,
2014). Even though these species may differ in some cases from the actual
compounds in the atmospheric PM due to thermal decomposition during analysis
(Lopez-Hilfiker et al., 2016), they serve as chemical tracers for the
atmospheric concentration of IEPOX-derived PM (Hu et al., 2015;
Isaacman-VanWertz et al., 2016). The analytical methods highlighted in
Fig. 1, including that of retrieving an “IEPOX-SOA factor” from mass spectral
analysis used
herein, can lead to over- and underestimated IEPOX-derived PM concentrations.
This uncertainty is represented in the figure by the brown dashed lines that
approximately but not exactly correspond to IEPOX-derived PM.
Under polluted conditions, the reactive sequence of isoprene and ultimately
PM production can become significantly altered (Fig. 1). NO concentrations
can be sufficiently high that ISOPOO radicals react almost entirely with NO
in place of HO2, thereby largely producing methacrolein (MACR) and
methyl vinyl ketone (MVK) in place of ISOPOOH (Liu et al., 2016a). As a
result, IEPOX production can be greatly decreased, ultimately reducing PM
production by IEPOX pathways. A minor channel along the NO pathway can still
produce IEPOX, although much less efficiently (Jacobs et al., 2014). Under
NO-dominant conditions, alternative pathways of PM production not involving
IEPOX can become active, though in lower yields. MACR can be oxidized to
produce peroxymethacrylic nitric anhydride (MPAN), which is a precursor to
methacrylic acid epoxide (MAE) and hydroxymethylmethyl-α-lactone
(HMML), and these compounds can undergo reactive uptake to produce PM
(Kjaergaard et al., 2012; Lin et al., 2013; Worton et al., 2013; Nguyen et
al., 2015). Glyoxal produced from isoprene oxidation can contribute to PM
production (Volkamer et al., 2007; Ervens and Volkamer, 2010; McNeill et al.,
2012; Marais et al., 2016).
Another possible mechanism affecting PM production by IEPOX uptake under
polluted conditions is altered particle composition, especially particle
acidity, largely driven by sulfate. Laboratory studies show that IEPOX uptake
increases with increasing acidity (Gaston et al., 2014; Kuwata et al., 2015;
Liu et al., 2015). A proposed reaction during uptake is the acid-catalyzed
ring opening of the IEPOX molecule (Surratt et al., 2010). The subsequent
particle-phase reactions include the addition of available nucleophiles, such
as water to produce tetrols or sulfate to produce organosulfates as well as
their oligomers (Surratt et al., 2010; Lin et al., 2014; Nguyen et al.,
2014). In support of this proposed mechanism, analyses by positive-matrix
factorization (PMF) of mass spectra collected in the southeastern USA
identified PMF factors associated with IEPOX-derived PM, and these factors
correlated positively with sulfate mass concentrations (Budisulistiorini et
al., 2013; Hu et al., 2015; Xu et al., 2015). In short, different regimes of
NO : HO2 concentration ratios and different possible PM compositions
between polluted and background conditions can lead to different product
distributions and different production rates of IEPOX-derived PM.
The extent to which pollution may shift the production pathways of
IEPOX-derived PM over tropical forests remains to be elucidated. The study
described herein is based on data sets collected in the wet season downwind
of Manaus, Brazil, during the Observations and Modeling of the Green
Ocean Amazon (GoAmazon2014/5) experiment (Martin et al., 2016a). The
research site was influenced at times and to variable extents by the
pollution outflow from the Manaus metropolitan area. Compared to the
background environment in Amazonia, the Manaus plume had high number
concentrations of particles and enhanced concentrations of pollutants,
including oxides of nitrogen and sulfate (Kuhn et al., 2010; Martin et al.,
2016a). The reactive gas-phase chemistry was strongly guided by the relative
mix of background and polluted air masses (Trebs et al., 2012; Liu et al.,
2016a). The analysis herein focuses on how the pollution perturbed
IEPOX-derived PM production relative to background conditions.
Methodology
Data sets were collected at the “T3” site (3.2133∘ S,
60.5987∘ W) located 70 km to the west of Manaus, Brazil, in central
Amazonia (Martin et al., 2016a). The site was situated in a pasture
(2.5 km × 2 km) surrounded by forest. The analysis herein focuses
on data sets collected during the wet season period of 1 February to
31 March 2014, corresponding to the first Intensive Operating Period (IOP1)
of the GoAmazon2014/5 experiment.
A High-Resolution Time-of-Flight Aerosol Mass Spectrometer
(HR-ToF-AMS, hereafter AMS; Aerodyne, Inc., Billerica,
Massachusetts, USA)
recorded the
primary data set of this study. The AMS provided quantitative bulk
characterization of the atmospheric PM. The design principles and
capabilities of this instrument are described in the literature (DeCarlo et
al., 2006; Canagaratna et al., 2007). The instrument was housed within a
temperature-controlled research container, and the inlet to the instrument
sampled from 5 m above ground level. Detailed aspects of AMS operation are
presented in the Supplement (Sect. S1). In brief, ambient measurements were
obtained every other 4 min. Organic, sulfate, ammonium, nitrate, and
chloride PM mass concentrations were obtained from “V-mode” data. The
choice of ions to fit was aided by the “W-mode” data, which were collected
for 1 day every 5 days. Data analysis was performed using SQUIRREL
(1.56D) and PIKA (1.14G) of the AMS software suite.
Positive-matrix factorization was applied to the time series of the organic
component of the high-resolution mass spectra (Ulbrich et al., 2009). The
present study focuses on one of the resolved statistical factors, referred to
as the IEPOX-SOA factor (Hu et al., 2015). Diagnostics of the PMF
analysis, especially as related to the resolved IEPOX-SOA factor, are
presented in the Supplement (Sect. S2). A separate account is forthcoming to
present the other PMF factors (de Sá, 2017). Herein, “factor
profile” and “factor loading” refer to the mathematical products of the
multivariate statistical analysis, whereas “mass spectrum” and “mass
concentration” refer to measurements.
Complementary to the AMS data sets, mass concentrations of molecular and
tracer organic species were measured using a semivolatile thermal desorption
aerosol gas chromatograph (SV-TAG) at a time resolution of 1 h. The
instrument collected gas and particle samples, followed by thermal
desorption, derivatization, and gas chromatography coupled to mass
spectrometry (Isaacman et al., 2014). A summary of operational details for
GoAmazon2014/5 is presented in the Supplement (Sect. S1), and the main
account is presented in Isaacman-VanWertz et al. (2016).
Additional data sets used in the analysis were collected at the T3 site by
instruments housed in the research container of the Mobile Aerosol Observing
System (MAOS) of the ARM Climate Research Facility (ACRF) operated by the USA
Department of Energy (Mather and Voyles, 2013; Martin et al., 2016a). A
temperature-controlled inlet was mounted at 10 m above ground level.
Measurements of nitrogen oxides were made using a chemiluminescence-based
instrument (Air Quality Design). The measured odd-nitrogen family
“NOy”, meaning NOx plus reservoir species, included NO,
NO2, HNO3, organonitrates, particle nitrate, and peroxyacetyl
nitrates. Further details of the NOy measurements are presented in the
Supplement (Sect. S1). Ozone concentrations were measured by an ozone
analyzer (Thermo Fisher, model 49i). Particle number concentrations were
measured by a condensation particle counter (TSI, model 3772). Meteorological
variables provided by the ARM Mobile Facility (AMF-1), which was also part of
the ACRF, included wind direction, solar irradiance, and precipitation rate.
Measurements of NOy and particle number concentration onboard the G-1
aircraft of the ARM Aerial Facility (AAF) were also used in the analysis
(Schmid et al., 2014; Martin et al., 2016a).
Results and discussion
The organization of the presentation herein is as follows. The factor profile
obtained from AMS PMF analysis is discussed (Sect. 3.1), a case study
comparing factor loading under background to polluted conditions is presented
(Sect. 3.2), the roles of sulfate (Sect. 3.3) and nitric oxide (Sect. 3.4) in
affecting factor loading are explored, and the influence of NO on production
and loss processes of IEPOX-derived PM is considered (Sect. 3.5).
Statistical IEPOX-SOA factor
Positive-matrix factorization was carried out on the time series of AMS
organic mass spectra. One statistical factor had a similar profile of peak
intensities as the IEPOX-SOA factor identified in other studies (Fig. S1 in the Supplement) (Robinson et al., 2011; Slowik et al., 2011; Budisulistiorini et al.,
2013, 2015; Chen et al., 2015; Xu et al., 2015). The Pearson correlation
coefficient R between this factor and the one obtained in the 2008 wet
season in central Amazonia as part of the AMAZE-08 experiment was 0.99 (Chen
et al., 2015). The ratio f of the factor loading to the mass concentration
of submicron organic PM for the present study was 0.17 ± 0.09
(mean ± standard deviation). The IEPOX-SOA factor has been identified
previously over the maritime tropical forest of Borneo (f=0.23) (Robinson
et al., 2011), in a rural area in Canada 70 km north of Toronto (f=0.17)
(Slowik et al., 2011), across several locations in the
southeastern USA in summertime (f=0.17 to 0.41) (Budisulistiorini et al., 2013, 2015, 2016; Hu et al., 2015; Xu et al., 2015; Marais et al., 2016), and in AMAZE-08 (f=0.34) (Chen et al., 2015). A further review of the ubiquity and
characteristics of the IEPOX-SOA factor is presented in Hu et al. (2015).
The IEPOX-SOA factor reported herein had prominent peaks at m/z 53.04 and
m/z 82.04 (Fig. S1). The ion at m/z 82.04,
corresponding to C5H6O+, has been attributed to 3-methylfuran
(3-MF). The thermal degradation of isoprene-derived PM upon mass spectral
analysis was suggested as the source of 3-MF (Robinson et al., 2011). Lin et
al. (2012) proposed that sequential dehydrations upon mass spectral analysis
of 3-methyltetrahydrofuran-3,4-diols, which are an identified component of
IEPOX-derived PM, can produce 3-MF. Other IEPOX-derived species as well as
non-IEPOX species might also contribute to the production of
C5H6O+ ions (Surratt et al., 2010; Hu et al., 2015; Liu et
al., 2016c).
Laboratory studies show that a mass spectrum that is a pattern of peak
intensities similar to that of the IEPOX-SOA factor is produced by the
uptake of IEPOX into aqueous acidic sulfate particles as well as by the
photooxidation of isoprene under HO2-dominant conditions in the presence
of acidic sulfate particles (Budisulistiorini et al., 2013; Nguyen et al.,
2014; Kuwata et al., 2015; Liu et al., 2015). The possibility of similar
uptake by a broader range of liquid media remains to be fully tested, such as
other acidic solutions as well as cloud waters. Compared to the laboratory
spectra of Liu et al. (2015), representing about 4 h of OH exposure at
atmospheric concentrations (1.7 × 106 molec cm-3), the
main difference was the relative intensity of the m/z 44 peak. For the
IEPOX-SOA factor of the present study, this peak was 4 times more intense
(Fig. S1), suggesting that the atmospheric PM was more oxidized. Hu et
al. (2016) showed that heterogeneous aging of IEPOX-SOA can result in
increased relative signal at m/z 44.
By contrast, laboratory studies show that a significantly different mass
spectrum from that of the IEPOX-SOA factor is obtained for PM produced from
isoprene photooxidation in the absence of aqueous particles (Krechmer et al.,
2015; Kuwata et al., 2015). Under these conditions, chemical pathways other
than IEPOX uptake into a liquid medium appear to be active, such as the
condensation of low-volatility, multifunctional compounds produced by
additional oxidation of ISOPOOH (Krechmer et al., 2015; Liu et al., 2016b;
Riva et al., 2016b). This non-IEPOX pathway, however, is not expected to
contribute a large fraction of the produced PM during the study period
because of the high relative humidity conditions in Amazonia and the prevalence of liquid
particles for the prevailing atmospheric conditions (Bateman et al., 2016; de
Sá, 2017).
The SV-TAG measurements of the concentrations of C5-alkene triols and
2-methyltetrols support the interpretation of the IEPOX-SOA factor as an
indicator that PM was being produced, at least in significant part, from the
reactive uptake of IEPOX (Claeys et al., 2004; Wang et al., 2005; Surratt et
al., 2010). The factor loading strongly correlated with the concentrations of
C5-alkene triols (R=0.96) and 2-methyltetrols (R=0.78) (Fig. 2).
These species have been associated with the IEPOX reaction pathway in several
laboratory studies (Surratt et al., 2010; Riedel et al., 2016). The R value
with respect to C5-alkene triols was independent of the
fpeak value of the PMF solution, demonstrating the robustness of
the relative time trend of factor loading even though the factor profile and
absolute loadings changed across fpeak values (Fig. S2d).
Scatter plot of the loading of the IEPOX-SOA factor derived from
analysis of the AMS data set and the mass concentrations of C5-alkene
triols and 2-methyltetrols measured by SV-TAG. All data collected during
the IOP1 period are included, meaning that the plotted data are not limited to
afternoon time periods.
The loading of the IEPOX-SOA factor may be an overestimate or an
underestimate of the atmospheric concentration of the IEPOX-derived PM
(Supplement, Sect. S2). The IEPOX-SOA factor can be understood as the net
result of (i) produced IEPOX-derived PM, (ii) minus that portion of the carbon
that gets further oxidized and mixed into other PMF factors, and (iii) plus
that portion of non-IEPOX-derived PM that gives rise to a similar AMS mass
spectral pattern as the IEPOX-derived PM (Supplement, Sect. S2). Processes of
type ii contribute to underestimates and processes of type iii lead to
overestimates when using IEPOX-SOA factor loading as a surrogate for
IEPOX-derived PM concentration. These uncertainties are qualitatively
represented in Fig. 1 by the brown dashed lines that enclose the fraction
of particle material statistically captured by the factor analysis. The
further analysis herein is based on using the loading of the IEPOX-SOA factor
as a scalar proxy for the mass concentration of IEPOX-derived PM in a sampled
air mass.
Background compared to polluted conditions
Under background conditions in the wet season, remote areas of the Amazon
forest constitute one of the least-polluted continental regions on Earth
(Martin et al., 2010a). Nitric oxide concentrations characteristic of
central Amazonia range from 20 to 70 ppt (Torres and Buchan, 1988; Bakwin et
al., 1990; Levine et al., 2015). Daytime maximum ozone concentrations are 10
to 15 ppb (Rummel et al., 2007). Sulfate mass concentrations associated with
in-basin processes are on average < 0.1 µg m-3, and total
background sulfate concentrations contributed by in- and out-of-basin
processes rarely exceed 0.5 µg m-3 (Andreae et al., 1990;
Chen et al., 2009).
In the wet season, Manaus emissions were the most important anthropogenic
influence on observations at the T3 research site (Martin et al., 2017). The
afternoons of 3 and 13 March 2014 are presented herein as representative
cases of background and polluted conditions, respectively. Both days were
sunny, and major precipitation events were absent. Particle number
concentrations measured onboard the G-1 aircraft within the atmospheric
boundary layer show the position of the pollution plume on these two
afternoons (Fig. 3). NOy concentrations measured during the same flight
are shown in Fig. S3. The visualization in Fig. 3 shows that on 3 March the
Manaus plume passed south of the T3 site. By comparison, on 13 March the
central portion of the plume passed over T3. Aircraft-based observations to
track the Manaus plume were available for 16 afternoons of the
2-month
study period.
Visualization of the Manaus pollution plume by plotting particle
number concentrations in the vertical axis. Observations took place on
flights from late morning to early afternoon on (a) 3 March 2014, 17:45–19:26 UTC,
and (b) 13 March 2014, 14:14–17:21 UTC. Local time is UTC - 4 h.
The red lines guide the eye through the central axis of the
plume. The direction and extent of the plume was observed by the G-1
aircraft within the atmospheric boundary layer downwind of Manaus. Measured
particle number concentrations are plotted on a vertical axis on top of an
image of land cover in the horizontal plane. Particle concentrations in the
center of the plume ranged from 10 000 to 25 000 cm-3 nearby Manaus.
Yellow pins indicate the locations of some of the GoAmazon2014/5 research
sites, including T3 (Martin et al., 2016a).
Case studies of (left) background and (right) polluted air masses
passing over T3 on afternoons of 3 and 13 March 2014. (a) Ozone, particle
number, and organic mass concentration. (b) IEPOX-SOA factor loading and the
ratio f of the factor loading to the organic PM concentration. (c) Sulfate
and NOy concentrations. (d) Wind direction, rain intensity, and solar
irradiance. Local time is UTC - 4 h. Time points of overflights at 500 m
by the G-1 research aircraft are marked by the dashed line (Martin et al.,
2016a).
Measurements at ground level at the T3 site are plotted in Fig. 4 for the
afternoons of 3 March (left panel) and 13 March (right panel). Based on wind
speeds, the research site was 4 to 6 h downwind of Manaus (Martin et al.,
2016a). Anthropogenic-biogenic interactions affecting the production of
IEPOX-derived PM were driven in large part by atmospheric photochemistry at
daybreak. Morning urban emissions followed by atmospheric processing arrived
at the T3 site during the local afternoon. The afternoon period, in addition
to the connection to the Manaus plume, was also characterized by reduced
variability in other possible confounding variables, such as temperature,
radiation, and relative humidity. Figure 4 shows that on the afternoon of
3 March ozone concentrations were below 10 ppb, particle number
concentrations were below 1000 cm-3, NOy concentrations were less
than 1 ppb, and sulfate concentrations were 0.3 to
0.4 µg m-3. Species concentrations were stable throughout the
afternoon. On 13 March, ozone concentrations exceeded 30 ppb for most of the
afternoon, particle concentrations reached 10 000 cm-3, NOy
concentrations consistently exceeded 1 ppb, and sulfate concentrations were
0.3 to 0.6 µg m-3. Concentrations fluctuated markedly
throughout the afternoon on 13 March, reflecting different levels of
pollution influence in the air passing over the T3 site during that period.
Elevated concentrations of ozone, particle number, and NOy were reliable
markers of pollution influence over the course of the study period
(Supplement, Sect. S3). Pollution was associated with stronger relative
enhancements in NOy concentrations than in sulfate concentrations
(Sect. 3.3 and 3.4). With respect to the IEPOX-SOA factor, Fig. 4 shows that
the absolute and relative loadings decreased for the polluted compared to
background conditions. Relative loadings are expressed by the ratio f of
IEPOX-SOA factor loading to the organic PM mass concentration. Decreased
absolute and relative factor loadings under polluted conditions, presented in
Fig. 4 as a case study, also characterized the data sets of the entire study
period. Other examples are presented in the Supplement (Fig. S4).
Sulfate as a driver of IEPOX-derived PM production
A scatter plot between sulfate mass concentrations and IEPOX-SOA factor
loadings for all afternoon periods is shown in Fig. 5a. Background and
polluted conditions are represented in the data set. For further
visualization, the data set was organized into six subsets based on sulfate
concentration. The medians and the means of the subsets are plotted in the
figure. The visualization shows that sulfate concentration served as a
first-order predictor of the IEPOX-SOA factor loading in central Amazonia in
the wet season. The explanation can be a combination of increased acidity,
greater reaction volume including by enhanced hygroscopic growth, and
possibly a nucleophilic role for sulfate (Xu et al., 2015; Marais et al.,
2016). An analysis of the relative importance of each is out of the scope of
the present study (Supplement, Sect. S4).
(a) Scatter plot of sulfate mass concentration and IEPOX-SOA
factor loading. A least-squares linear fit is represented by the dashed line
(R2=0.37). The data set was collected into six subsets based on
sulfate concentration to calculate statistics. Medians (squares) and means
(diamonds) of each subset are plotted. Whiskers on the medians represent the
interquartile ranges. (b) Probability density function of sulfate mass
concentration at the background site T0t (TT34) north of Manaus in the
wet season of 2008 (Chen et al., 2009; Martin et al., 2010b,
2016b), at the background site T0a (ATTO) northeast of Manaus in the wet
season of 2015 (Andreae et al., 2015), and at T3 during the wet season of
2014 (IOP1). The plotted data sets were recorded during local afternoons
(12:00–16:00 local time; 16:00–20:00 UTC). Means (diamonds), medians
(squares), and interquartile range (whiskers) are shown for the probability
density functions.
For Fig. 5a, the coefficient R2 of determination between sulfate mass
concentration and factor loading was 0.37, meaning that 37 % of the
variance of the IEPOX-SOA factor loading was explained by sulfate mass
concentration. As a point of comparison, R2 varied between 0.4 and 0.6
for observations in the southeastern USA, which seasonally is a region of
high isoprene emissions (Budisulistiorini et al., 2013, 2015; Hu et al., 2015; Xu et al., 2015). A chemical transport model that
predicted IEPOX-derived PM mass concentration for the southeastern USA
obtained R2 of 0.4 for the relationship to predicted sulfate mass
concentration (Marais et al., 2016). The model attributed the correlation to
the acidity and particle volume provided by sulfate, both of which favored
IEPOX uptake. Central Amazonia and the southeastern USA differ considerably
in terms of meteorology, chemistry, and levels of regional pollution, yet
they have in common an important role of sulfate concentration as a predictor
of IEPOX-derived PM concentration, even as the sulfate concentrations
themselves differ by an order of magnitude. Sulfate concentrations typically
had an interquartile range of [1.5, 3.0] µg m-3 in the
studies in the southeastern USA, which can be compared to a range of [0.11,
0.36] µg m-3 under background conditions during the wet
season in central Amazonia.
A key difference between the southeastern USA and central Amazonia is the
role of sulfate concentration as a clear or ambiguous indicator,
respectively, of urban influence. For the relatively low sulfate mass
concentrations (< 0.5 µg m-3) characteristic of the study
period, background air in central Amazonia contributed significantly to the
variability in sulfate concentration measured at the T3 site. Background
concentrations of sulfate in Amazonia, distinguished from sulfate tied to the
urban Manaus plume, originated from in-basin emissions of gas-phase
precursors such as dimethyl sulfide (DMS) and hydrogen sulfide (H2S)
from the forest as well as from out-of-basin marine emissions from the
Atlantic Ocean (Andreae et al., 1990; Chen et al., 2009; Martin et al.,
2010a). In the wet season, biomass burning from Africa and to a lesser extent
from South America also episodically contributed significantly to sulfate
concentrations in the Manaus region. In addition, emissions from large cities
on the eastern coast of Brazil were important at times when rare
meteorological events shifted the northeasterlies typical of the wet season
to easterlies (Martin et al., 2017). Manaus contributions to sulfate mass
concentrations in an air mass were in addition to these various background
sources.
The relative importance of Manaus contributions to the sulfate concentrations
in the air masses that passed over T3 was assessed by comparison of the
probability density function of sulfate concentration at T3 to those of sites
upwind of Manaus (Fig. 5b). The distributions of the two upwind sites had a
central tendency of 0.05 to 0.3 µg m-3, suggesting the range
of natural concentrations, and a right-side skewness up to
0.6 µg m-3, suggesting the importance of episodic long-range
transport (Chen et al., 2009). The figure shows that the distribution at T3
did not differ greatly from those of the upwind sites even though the air
masses over T3 regularly transported Manaus pollution. The implication is
that Manaus sulfate sources, whether primary or secondary, had small
contributions relative to background sources when averaged over time. In
short, elevated sulfate concentrations on any one afternoon at the T3 site
might have arisen because of elevated background concentrations on that day
rather than the influence of the Manaus pollution plume. The implications are
that (i) sulfate concentration was an ambiguous indicator of urban influence
at the T3 site and (ii) increases in sulfate concentrations in pollution
events were moderate relative to background concentrations.
Parameters associated with the NOy groupings of Fig. 6. Listed
are the NOy concentrations and the parameters for least-squares linear
fits to each group. R2 represents the coefficient of determination.
Group
NOy range (ppb)
Fit slope
Fit intercept
Fit R2
1
> 0.66
2.16
-0.13
0.75
2
0.66–0.92
1.48
-0.04
0.64
3
0.92–1.55
0.78
0.06
0.24
4
1.55–2.45
0.71
-0.01
0.44
5
> 2.45
0.55
-0.02
0.62
NO as a modulator of IEPOX-derived PM production
In the transport from Manaus to the T3 research site, NO concentration was
not conserved, in part because of reactions with ozone and organic peroxy
radicals (Martin et al., 2017). In this case, the instantaneous NO
concentrations measured at the T3 site did not directly provide information
about the fate of ISOPOO radicals along the transport time of 4 to 6 h from
Manaus to the T3 site. The collective contributions of NO, NO2, and
their oxidation products were, however, reflected in measurements of NOy
concentrations at the T3 site. The NOy family is expected to have a
longer lifetime than the transport time from Manaus to the T3 site (Romer et
al., 2016). The NOy concentration measured at T3 therefore served as a
surrogate for the integrated exposure of the air mass to NO chemistry between
Manaus and T3 (Liu et al., 2016a).
Unlike the ambiguity associated with the sulfate concentration, an elevated
NOy concentration served as a clear indicator of anthropogenic influence
in an air mass passing over the T3 site. For background conditions over the
forest, NOy originates from NO emitted from soils and other natural
sources such as lightning (Bakwin et al., 1990; Jacob and Wofsy, 1990). The
probability density function of NOy concentration under background
conditions in the wet season of the central Amazon basin is shown in Fig. 6b
(Bakwin et al., 1990). The distribution for measurements at T3 is also shown.
Relative to the narrow distribution around 0.5 ppb for background
conditions, there is high-side skewness extending up to 4 ppb for the T3
measurements, indicating the clear influence of Manaus emissions on NOy
concentrations.
(a) Scatter plot of sulfate mass concentration and IEPOX-SOA
factor loading for local afternoon (12:00–16:00 local time; 16:00–20:00 UTC). The data sets were collected into five subsets, colored and labeled 1
to 5, based on NOy concentration. Table 1 presents the parameters of
the five least-squares linear fits represented by the colored lines in the
figure. (b) Probability density function of NOy concentration at a
background site nearby Manaus in the wet season of 1987 (Bakwin et al.,
1990) and at T3 during the wet season of 2014 (IOP1) (afternoon data). Means
(diamonds), medians (squares), and interquartile range (whiskers) are shown
for the probability density functions. Additional analysis of (a) is
presented in the Supplement (Sect. S5) related to Fig. S5.
NOy concentration was incorporated into the analysis by segregation of
the data set of Fig. 5a into five subsets (Supplement, Sect. S5). Linear fits
to the NOy-segregated data subsets are plotted in Fig. 6a. Each subset
is represented by a different color. Parameter values of the associated fits
are listed in Table 1. In conjunction with sulfate concentration, the
visualization presented in Fig. 6a shows that NOy concentration
further explained the variability in IEPOX-SOA factor loadings. The R2
values, representing the extent to which sulfate was able to explain
variability in IEPOX-SOA factor loading once isolated for NOy
concentration, were higher for the data subsets with lower and higher
extremes of NOy concentrations (Table 1). These conditions represent the
limiting cases of fully background conditions for the former and the
strongest effects of Manaus pollution for the latter. By comparison,
intermediate NOy concentrations could arise from air masses that mixed
together background air with Manaus pollution during the transport to T3
(e.g., by entrainment) and thus represent complex processing. Single or
multiple mixing points could occur anywhere along the path from Manaus to T3,
thus introducing variability into the effective photochemical age of the air
mass arriving at T3 and resulting in lower R2 values for intermediate
NOy concentrations. A caveat is that this explanation assumes that NO
emissions from Manaus had low day-to-day variability.
In relation to the influence of Manaus pollution, sulfate concentration was
affected by a mixture of background and urban sources (see discussion in
Sect. 3.3) whereas NOy concentration largely had urban sources (see
Figs. 5b and 6b). As an approximation to keeping the sulfate concentration
constant and thus focusing on the role of NO in the urban pollution, the
visualization of the dependence of IEPOX-SOA factor loading on NOy
concentration was further refined by taking data subsets segregated by low
(< 0.1 µg m-3) and high (> 0.3 µg m-3)
sulfate concentrations. Figure 7a, b, and c show the factor loading, organic
PM mass concentration, and the ratio f of the IEPOX-SOA factor loading to
the organic PM mass concentration, respectively, plotted against NOy
concentration for low and high sulfate concentrations.
Figure 7a shows that for both low and high sulfate concentrations an increase
in NOy concentration from background to polluted concentrations was
associated with a decrease in the IEPOX-SOA factor loading by two to three
times. For low sulfate concentration, the interquartile range of the factor
loading decreased from [0.037, 0.093] to [0.022,
0.039] µg m-3 for an increase in NOy concentration from
0.5 to 2 pbb. For high sulfate concentration, the factor loading decreased
from [0.57, 0.95] to [0.21, 0.35] µg m-3 for the same
transition in NOy concentration. The greatest changes in factor loading
were in the region of 1 ppb NOy. This region of greatest sensitivity
coincided with the transition from background to polluted conditions.
For the same time period of these PM analyses of IEPOX-SOA factor loading,
Liu et al. (2016a) observed a shift in dominant isoprene gas-phase products
from ISOPOOH to MVK/MACR across the transition in NOy concentration. Liu
et al. (2016a) further simulated the dependence on NO concentration of the
ratio of the production rate of ISOPOOH to that of MVK + MACR. The
highest ratios (0.6 to 0.9) were obtained for background concentrations of
NOy. The calculated HO2 concentration was
< 4 × 108 cm-3 (0.016 ppb). The simulated transition
for the dominant fate of the ISOPOO radicals occurred for an NO concentration
of < 0.05 ppb.
Dependence on NOy concentration of (a) IEPOX-SOA factor
loading, (b) organic mass concentration, and (c) the ratio f of the IEPOX-SOA
factor loading to the organic PM concentration. Data are segregated by low
(< 0.1 µg m-3) and high (> 0.3 µg m-3) sulfate mass concentration and grouped into five levels of
NOy concentration (Fig. 6). Squares represent medians of each group.
Interquartile ranges are represented by whiskers along the abscissa and
shading along the ordinate. The plotted data sets were recorded during local
afternoon (12:00–16:00 local time; 16:00–20:00 UTC).
Figure 7b shows that for both low and high sulfate concentrations the organic
PM mass concentration Morg and the IEPOX-SOA factor loading had
opposite trends for low compared to intermediate NOy concentrations,
even though the trend in Morg was less steep. The factor loadings
decreased by 60 % whereas the Morg increased by 25 % for
0.5 to 2 ppb NOy (Fig. 7a and b). Increases in Morg can
include contributions from secondary PM produced by enhanced concentrations
of hydroxyl radicals and ozone in the pollution plume as well as from primary
PM emitted from the Manaus urban region (Martin et al., 2017; de Sá,
2017). For higher NOy concentrations (> 2 ppb), however, Fig. 7b
shows that Morg decreased after a peak value, approaching values
close to background under the most polluted conditions. The chemistry can
become sufficiently shifted that more-volatile gas-phase products can be
produced (Pandis et al., 1991; Kroll et al., 2005; Carlton et al., 2009). In
addition, hydroxyl radical concentrations can also decrease because of
titration by NO2 (Valin et al., 2013; Rohrer et al., 2014). An increase
in total organic mass concentration could possibly contribute to a decrease
in IEPOX-derived PM production by kinetically limiting the uptake of IEPOX
(Gaston et al., 2014; Lin et al., 2014; Riva et al., 2016a). The dominant
effect of the urban plume, however, seems to be that of shifting the fate of
ISOPOO radicals through the increase in NO, thereby significantly decreasing
the production of ISOPOOH (Liu et al., 2016a) (Sect. 3.5).
The combined trends of Fig. 7a and b for increasing NOy are represented
in Fig. 7c as the ratio f. The figure shows that f decreased for
increasing NOy concentration for both low and high sulfate
concentrations. The greatest decrease occurred across the range of NOy
concentrations that represented the shift from background to polluted
conditions. For low sulfate concentration, the interquartile range of f
decreased from [0.09, 0.18] to [0.04, 0.09] for an increase in NOy
concentration from 0.5 to 2 ppb. These ranges shifted to [0.35, 0.40] and
[0.07, 0.18] for high sulfate concentration. As a limiting statement, for the
most favorable conditions with respect to the production of IEPOX-derived PM
in central Amazonia (i.e., lowest NOy and highest sulfate), f exceeded
0.40 at 25 % frequency. The implication is that at all times significant
additional pathways for PM production were active. This conclusion is subject
to the accuracy of the IEPOX-SOA factor loading as a scalar proxy of
IEPOX-derived PM concentration (see discussion of Fig. 1 in Sect. 3.1). The
magnitude of the decrease in f for high sulfate concentrations suggests
that IEPOX-derived PM shifted from being a major to a minor component of the
PM. Taken together, the results shown in Fig. 7 demonstrate how urban
pollution affected the production and composition of regional IEPOX-derived
PM.
The data sets presented in Figs. 5, 6, and 7 lead to the conclusion that the
additional NO concentrations contributed by Manaus emissions typically
suppress the production of IEPOX-derived PM to a greater extent than the
additional sulfate concentrations enhance it. Figure 8 presents a systematic
visualization. The factor loadings at T3 are represented as contours for axes
of sulfate and NOy concentrations. Higher factor loadings are favored
for higher sulfate and lower NOy concentrations. Factor loadings are
most sensitive to changing concentration in the high-sulfate, low-NOy
region. The gray dashed line in Fig. 8 represents a qualitative divisor
between domains of typical background and polluted conditions downwind of
Manaus.
Contours of IEPOX-SOA factor loading for sulfate and NOy
concentrations. The plotted data were recorded during local afternoon
(12:00–16:00 local time; 16:00–20:00 UTC). Typical transition between
regimes of background and polluted conditions for the region downwind of
Manaus are approximately represented by the dashed gray line.
Descriptions and units of symbols in the model.
Symbol
Description
Unit
M
Mass concentration of IEPOX-derived PM
µg m-3
t
Time
h
kP
Zero-order rate coefficient for production under background conditions
µg m-3 h-1
kL
First-order rate coefficient for loss under background conditions
h-1
α
Multiplicative factor representing the effects of Manaus pollution on rate coefficients
τ
Characteristic time of a process (e.g., production, loss, or transport)
h
Subscript tr
Refers to transport
Subscript bg
Refers to background conditions
Subscript pol
Refers to polluted conditions
Subscript 0
Refers to an initial state (i.e., just upwind of Manaus)
Subscript P
Refers to production processes
Subscript L
Refers to loss processes
Influence of NO on production and loss processes of IEPOX-derived
PM
Elevated NO concentrations may affect both the production and loss processes of
IEPOX-derived PM. On the one hand, production may be reduced because of
increased scavenging of ISOPOO by NO, thus obviating production of IEPOX and
consequently of IEPOX-derived PM. Production may also be reduced because of
more rapid gas-phase loss of IEPOX in response to elevated OH and O3
concentrations. On the other hand, loss of IEPOX-derived PM may be enhanced
due to faster processing of its characteristic compounds by the elevated
oxidant concentrations.
A Lagrangian model is employed to help delineate the relative importance of
reduced production compared to enhanced loss on the observed IEPOX-derived PM
concentrations. The model is initialized by background air that passes over
Manaus in the mid-morning. The evolution of IEPOX-derived PM in that air mass
is modeled under either polluted or background conditions for arrival at the
T3 site in the afternoon. The governing differential equation of the model
represents the sum of production and loss processes affecting the
concentrations of IEPOX-derived PM, as follows:
dMdt=-αLkLM+αPkP,
where M designates the IEPOX-derived PM mass concentration, t designates
time, and the first and second terms on the right-hand side represent loss
and production processes, respectively. Table 2 lists other symbol
definitions and units.
The analytic solution of Eq. (1) for time t is presented in the Supplement
(Sect. S6). From this solution, characteristic times τ for production
and loss processes for polluted compared to background conditions are as
follows: τP,pol=M0/(αPkP),
τP,bg=M0/kP, τL,pol=1/(αLkL), and τL,bg=1/kL (Supplement, Sect. S6). The term M0 represents the
IEPOX-derived PM mass concentration just upwind of Manaus. Under background
conditions, the enhancement factors αL and
αP are unity by definition. Under polluted conditions,
αL=2 and αP=0.1 to reflect enhanced loss
and decreased production, respectively. Further descriptions of the model and
assumptions are presented in the Supplement (Sect. S6).
Modeled IEPOX-derived PM mass concentrations Mpol and
Mbg at the T3 site under polluted compared to background
conditions. The ratio ξ of concentrations (i.e.,
Mpol/Mbg) is also plotted. Panels (a, b) correspond
to Case 1 and Case 2 listed in Table 4.
Gray shading indicates the range of observed values of ξ across low and
high sulfate concentrations. Dashed lines indicate the intersection of
modeled and observed values of ξ and the corresponding constrained
values of kP or kL along the abscissa. Labels above
the double-headed arrow in (b) correspond to characteristic times (i.e.,
kL-1). The dashed black arrow in (b) communicates that
the observed values of ξ provide no constraint on the lower limit of
kL.
Interquartile intervals of IEPOX-SOA factor loadings observed for
background and polluted conditions. Background and polluted conditions
correspond to approximately 0.5 and 2 ppb of NOy, respectively. The
table also lists the resulting ratio ξ of the median factor loading
under polluted compared to background conditions.
Loading (µg m-3)
Loading (µg m-3)
Ratio
for background
for polluted
ξ
conditions
conditions
Low sulfate
High sulfate
Low sulfate
High sulfate
Low sulfate
High sulfate
IEPOX-SOA factor
[0.037, 0.093]
[0.57, 0.95]
[0.022, 0.039]
[0.21, 0.35]
0.47
0.35
The analysis strategy is to compare τP and τL
to the transport time τtr under polluted and background
conditions. In this way, the analysis assesses the relative importance of altered production and loss
processes for IEPOX-derived PM from Manaus to T3. The statistical mode value
for τtr of 4 h based on trajectory analysis is used in the
model (Martin et al., 2016a). Intervals for the characteristic times
τP and τL are constrained by the T3 afternoon
data sets. Concentration ratios ξ, defined as ξ=Mpol/Mbg, are used to constrain the model (Table 3). The quantities
Mpol and Mbg denote M(t=τtr),
meaning the mass concentration at T3 under polluted or background conditions,
respectively. The use of the ratio quantity ξ in the analysis, rather
than absolute concentrations, provides increased robustness because of low
variability in ξ across the observed range of sulfate concentrations,
even as Mpol and Mbg vary greatly (Table 3). The
possible impacts of over- or underestimates of IEPOX-derived PM mass
concentration, as a consequence of using IEPOX-SOA factor loading as a
surrogate, are also mitigated by the use of ξ.
Two cases of the model (1 and 2) are presented, respectively focusing on
constraining kP or kL and consequently
τP or τL (Table 4). The results for Case 1 of
the analysis are shown in Fig. 9a. The value of kP is varied from
0 to 0.2 µg m-3 h-1, and the other model parameters
are held constant. The loss rate coefficient kL is fixed at
0.015 h-1, corresponding to a characteristic time of 2.8 days
(Supplement, Sect. S6). Based on observed values of ξ (gray shaded area
in Fig. 9a), an interval for kP of [0.07,
0.13] µg m-3 h-1 is obtained, as indicated by the
vertical dashed lines. Across this interval, Mbg and
Mpol vary from 0.49 to 0.72 µg m-3 and 0.23 to
0.25 µg m-3, respectively, which are consistent with the
observed IEPOX-SOA factor loadings (Table 3). The modeled production times
have intervals of [1.8, 3.3] h for τP,bg and [18, 33] h
for τP,pol.
Parameter values and initial conditions used for the model cases.
Descriptions and units are listed in Table 2. The M0 value is based on
the mean IEPOX-SOA factor loading that was measured from 14:00 to 16:00 UTC
(10:00–12:00 local time) at a regional background site during 2 months of
the wet season in 2008 (Chen et al., 2015). For comparison, a similar value
of 0.19 µg m-3 was observed during the present study as the
mean at the T3 site for NOy < 1 ppb (14:00–16:00 UTC).
Model case
Parameter values
Initial condition
kP
kL
αP
αL
M0
1. Vary kP
0 to 0.2
0.018
0.1
3
0.23
2. Vary kL
0.065
0.001 to 1
0.1
3
0.23
Case 2 of the analysis evaluates constraints on the loss rate coefficient
kL, and results are shown in Fig. 9b. Loss processes can include
chemistry, such as heterogeneous oxidation or other in-particle reactions
that reduce the IEPOX-SOA factor loading, as well as physical mechanisms,
such as particle deposition and particle dilution by entrainment that reduce
mass concentrations of IEPOX-derived PM (Supplement, Sect. S6). The value of
kL is varied over 3 orders of magnitude, representing
characteristic times of hours to weeks, and the other model parameters are
held constant (Table 4). The production rate coefficient kP is
fixed at 0.10 µg m-3 h-1, corresponding to the interval
midpoint of Case 1. The observed values of ξ (gray shaded area) in
intersection with the modeled values of ξ imply an upper limit on
kL at 0.043 h-1, corresponding to characteristic times of a
day to weeks (Fig. 9b). Correspondingly, τL,bg > 24 h
under background conditions, and τL,pol > 12 h under
polluted conditions.
The analyses of cases 1 and 2 constrain the values of τP,pol,
τP,bg, τL,pol, and τL,bg based on the
observed values of ξ. The lower limits of the characteristic times for
loss, meaning τL,bg > 24 h and τL,pol
> 12 h, are considerably longer than the transport time of 4 h under
both background and polluted conditions. Enhanced loss, therefore, does not
explain alone the observed values of ξ. By comparison, the observed
values of ξ imply a shift in the characteristic time for production from
[1.8, 3.3] h under background conditions to [18, 33] h under pollution
conditions. The shift in timescale is significant in light of the transport
time of 4 h. Therefore, reduced production, rather than enhanced loss, is
consistent with the lower IEPOX-derived PM concentrations under polluted
conditions. A few afternoon hours of altered isoprene chemistry is sufficient
to significantly shift the atmospheric concentration of IEPOX-derived PM.
Summary and conclusions
The influence of anthropogenic emissions on the production of organic
particulate matter from isoprene epoxydiols was studied during the wet season
of the tropical forest in central Amazonia. The IEPOX-derived PM
concentration at the T3 site, as indicated by the IEPOX-SOA factor loading,
was lower under polluted compared to background conditions. Sulfate
concentration was an important first-order predictor of the IEPOX-SOA factor
loading, corroborating the understanding of the role of sulfate in the
production of IEPOX-derived PM that has been developed in laboratory studies
as well as in investigations in the southeastern USA (Surratt et al., 2007b;
Budisulistiorini et al., 2013, 2015; Hu et al.,
2015; Kuwata et al., 2015; Xu et al., 2015). Unlike the southeastern USA,
however, where anthropogenic influences dominated variability in sulfate
concentrations, contributions by the Manaus urban region to sulfate
concentrations were of approximately equal magnitude to the background
variability in central Amazonia. By comparison, Manaus urban emissions of NO
dominated over background concentrations, and the NOy concentration
measured 4 to 6 h downwind of Manaus at the T3 site was an important
predictor of the IEPOX-SOA factor loading. In net effect, the suppression of
IEPOX production because of elevated NO concentrations in the pollution plume
dominated over any enhancements in IEPOX uptake because of greater sulfate
concentrations.
The dependence of the IEPOX-SOA factor loadings on both sulfate and NOy
concentrations, as shown in Fig. 8, suggests that altered net anthropogenic
effects may be expected for different geographic regions, even within
Amazonia, and different time periods, such as the wet and dry seasons. The T3
site experienced a wide range of NOy concentrations, allowing for the
systematic demonstration of the dependence of IEPOX-derived PM concentrations
on NOy concentrations. The results show that the transition in isoprene
photochemistry related to the production of IEPOX-derived PM is most
sensitive precisely at the transition between background and polluted
conditions, around 1 ppb of NOy, at least for central Amazonia in the
wet season. These findings suggest that the fraction of PM derived from IEPOX
might be lower and have lower variability for other geographic regions
characterized by higher NOy baseline concentrations (e.g., upward of 1
to 2 ppb). For regions further downwind of the urban center, the effects of
the plume are expected to phase out both due to dilution and to consumption
of NO, and a gradual transition to background chemistry is expected to take
place. Adequately representing background conditions and the transition to
polluted conditions within models, including the dependence of the production
of IEPOX-derived PM not only on sulfate but also on NO concentration, is thus
important for making accurate predictions of PM concentrations, both in
Amazonia and around the globe.
The findings herein can be considered in the context of Amazonia in
transition (Davidson et al., 2012). In the past 50 years, the metropolitan
area of Manaus, today at more than 2 million inhabitants, has experienced
rapid economic and population growth (Martin et al., 2016a). Changes in the
fuel matrix, such as the ongoing shift from high-sulfur to low-sulfur oil in
the vehicle fleet as well as from fuel oil to natural gas in many power
plants (Medeiros et al., 2017), are changing the composition of the Manaus
pollution plume. Based on the findings presented herein, a reduction in
sulfate sources from Manaus, whether primary or secondary, would not be
expected to considerably affect the mass concentration of IEPOX-derived
species in forest regions affected by the plume. Background sources
independent of Manaus appear sufficient to sustain sulfate concentrations
regionally. In the absence of pollution-control
technologies, however, NO emissions can be expected to increase in coming years due to
the development of more efficient (i.e., higher temperature) sources of
electricity associated with the development of natural gas resources in the
basin, as well as from growth in transportation associated with increased
population. Increased NO concentrations can be expected to reduce the mass
concentration of IEPOX-derived species in forest regions affected by the
plume. Changes in the atmospheric particle population can have follow-up
effects on cloud type, duration, and rainfall (Pöschl et al., 2010). In
addition to PM derived from IEPOX as discussed herein, a better understanding
of other pathways that also contribute to organic PM, as well as possible
changes to those pathways with increasing pollution in the region, warrants
further study so as to achieve sufficient knowledge for decision-making
related to air quality and climate in Amazonia.