-
Soot superaggregates from flamingwildfires and their direct
radiative forcingRajan K. Chakrabarty1,2, Nicholas D. Beres2, Hans
Moosmuller2, Swarup China3, Claudio Mazzoleni3,Manvendra K. Dubey4,
Li Liu5 & Michael I. Mishchenko5
1Department of Energy, Environmental & Chemical Engineering,
Washington University in St. Louis, St. Louis, MO 63130,
USA,2Desert Research Institute, Nevada System of Higher Education,
Reno, Nevada, USA, 3Atmospheric Sciences Program,
MichiganTechnological University, Houghton, Michigan, USA, 4Earth
System Observations, Los Alamos National Laboratory, Los Alamos,New
Mexico, USA, 5NASA Goddard Institute for Space Studies, New York,
NY 10025, USA.
Wildfires contribute significantly to global soot emissions, yet
their aerosol formation mechanisms andresulting particle properties
are poorly understood and parameterized in climate models. The
conventionalview holds that soot is formed via the cluster-dilute
aggregation mechanism in wildfires and emitted asaggregates with
fractal dimension Df < 1.8 mobility diameter Dm # 1 mm, and
aerodynamic diameter Da #300 nm. Here we report the ubiquitous
presence of soot superaggregates (SAs) in the outflow from a
majorwildfire in India. SAs are porous, low-density aggregates of
cluster-dilute aggregates with characteristicDf < 2.6, Dm . 1
mm, and Da # 300 nm that form via the cluster-dense aggregation
mechanism. We presentadditional observations of soot SAs in
wildfire smoke-laden air masses over Northern California,
NewMexico, and Mexico City. We estimate that SAs contribute, per
unit optical depth, up to 35% lessatmospheric warming than
freshly-emitted (Df < 1.8) aggregates, and
-
temperature). Exceptions have been the aircraft sampling
studiesconducted during the 1990s in the over-fire regions of
flaming forestfires in Brazil and southern Africa17,18.
Observations of unusuallylarge soot aggregates were made from these
fires, but the investiga-tors failed to distinguish the
microphysical properties of these part-icles from conventional
sub-micron soot aggregates. As a result,these unique observations
have gone unnoticed, and there has beenno follow-up investigation
conducted on the formation mechanismand frequency of occurrence of
these unusually large soot aggregates,their microphysical
properties, and their potential impact on radi-ative forcing and
health.
Here, we investigate particles contained in the
flaming-phaseplumes of the Nagarhole National Forest fire (NNFF)24
inKarnataka (India) and find the ubiquitous occurrence of
superag-gregates (SAs), a hitherto unrecognized form of soot
distinct fromconventional sub-micron aggregates. We report
additional obser-vation of these SAs in wildfire smoke-laden air
masses overSacramento (Northern California, USA), Los Alamos
(NewMexico, USA), and the Mexico City metropolitan area
(Mexico),respectively. Based on the unique morphological properties
of SAs,we discuss their possible formation mechanism and their
potentialimpact on human health. We also compute numerically-exact
opticalproperties of these particles and compare them with those of
sub-micron size soot particles. We make use of the optical
properties tocalculate direct radiative forcing efficiencies of SAs
at the top of theatmosphere and discuss their net warming or
cooling of the atmo-sphere. Finally, we address the need for future
research to betterunderstand and characterize the detection and
atmospheric proces-sing of soot SAs for quantitatively estimating
their impact on climateand health.
ResultsWe collected aerosol samples for scanning electron
microscopy (SEM)analysis downwind of the NNFF over the Indian Ocean
at the MaldivesClimate Observatory on Hanimaadhoo Island (MCOH)
(6.78u N,73.18u E). The NNFF, which lasted for a week beginning
February27, 2012, burned approximately 35 km2 of dry deciduous
forest con-taining dry bamboo and teak trees. Dense smoke from
intense flamingcombustion was reported, with the event turning into
a firestormwithin a day24. The MCOH aerosol number concentration
increasedfrom about 800 to 3000 cm23 during this period. The months
ofNovember through May constitute the dry season in South Asia,
whenlow-level flow brings a polluted air mass from Asia to the
IndianOcean25. Ensemble back-trajectory analyses (Fig. 1a; Fig. S1
in supple-mentary information) coupled with satellite imagery and
the CloudAerosol Lidar and Infrared Pathfinder Satellite (CALIPSO)
measure-ments (Fig. 1 b and c) show a low-level polluted air
massbetween 1to 3 km above sea leveltransported from the forest
fire site flowingsouthwest over the Indian Ocean. Gas
chromatography interfaced withmass spectrometry analyses of aerosol
samples, revealed the presenceof levoglucosana molecular marker for
biomass burning emis-sions26in trace amounts (about 0.09
ng/m3).
To investigate how commonly these SAs occur in different
geo-graphical locations and atmospheric conditions, we sampled
aerosolscontained in wildfire smoke-laden air masses over
Sacramento dur-ing the CARES (Carbonaceous Aerosol and Radiative
EffectsStudy)27 in June 2010 and over Mexico City as part of
theMILAGRO (Megacity Initiative: Local And Global
ResearchObservations)28,29 study during March-June 2006. Finally,
in 2011at Los Alamos, we sampled the downwind plumes of the
LasConchas wildfire19, the second largest wildfire in the states
history.
Figure 1 | The 2012 Nagarhole forest wildfire smoke plume
transport. (A) Average of NOAA HYSPLIT ensemble trajectories ending
at the MaldivesClimate ObservatoryHanimaadhoo, Maldives (MCOH) on
01, 02, and 03 March 2012. (Image created using Adobe PhotoshopTM);
(B) Visible imagery of
the Indian Peninsula from the MODIS sensor aboard the Terra
satellite for 28 February 2012. (Image obtained from NASA Near
Real-time (NRT) data
archive); (C) 532 nm backscatter return signal from the CALIOP
Lidar aboard the CALIPSO satellite showing vertical distribution of
aerosols (Image
obtained from NASA CALIPSO data archive). The color scale on the
right indicates the strength of the LIDAR return signal: boundary
layer clouds usually
show up as grey or white; cirrus clouds range from yellow to
grey; and aerosols show up as green, yellow, and red (indicating
low, medium, and high
loadings, respectively).
www.nature.com/scientificreports
SCIENTIFIC REPORTS | 4 : 5508 | DOI: 10.1038/srep05508 2
-
Morphological properties of soot superaggregates. We measuredthe
structural and fractal properties of individual
carbonaceousparticles collected from smoke plumes over the four
locationsusing established image analysis routines5. The fractal
properties ofindividual particles were quantified using the
perimeter andensemble methods3032. Electron micrographs of typical
SAsobserved at the four sampling sites are shown in Fig. 2 and 3.
TheSAs consisted of aggregates of sub-micrometer size,
cluster-dilutesoot aggregates with characteristic Df 5 1.9 6 0.2
(Fig. 2c). Themean maximum length scales of SAs were between 10 and
20 mm,and SAs had distinct Df 5 2.6 6 0.1.
Our analysis showed that a typical SA consisted of around
3000monomers, after accounting for apparent monomer overlap, which
isparameterized by a power-law factor 1.0933. Observation of Df 5
1.9sub-micron aggregates within individual SAs confirms that the
SAswere formed via percolation of these aggregates in the fires.
Highmagnification images (Fig. 2d) of NNFF revealed minimal coating
ofcondensed organic matter on the monomers. This suggested
thatthese particles were formed under near unity net equivalence
ratio34
and resisted atmospheric processing during long-range
transport11.Alternatively, it could be that there was not enough
condensableorganic matter available in the flaming fire plumes of
NNFF (seetable S1 in supplementary information) to coat the
SAs.
We analyzed 69 individual carbonaceous particles collected onSEM
filters from the NNFF. Approximately 99% and 75% of theparticle
mass and number, respectively, were soot SAs9,31,32,35, withthe
remaining being aggregates (Fig. 4). We probed the
elementalcomposition of the particles using energy dispersive X-ray
spectro-scopy (EDX), finding carbon and oxygen to be their primary
con-stituents. No tar balls22,36 or particles with inorganic
inclusions wereobserved. We analyzed 580 particles from the CARES
campaign andfound approximately 16% of the total aerosol particle
number wasSAs. The remaining 57% and 27% of particles were
externally sub-micron soot (occurring as bare and mixed with
organic carbon) anddust particles, respectively. The SAs observed
in samples fromMILAGRO and the Las Conchas fires were less than 1%
in number.Aerosol types in these two locations were mostly organic
carbon and
tar balls19,20,28,29,37, suggesting dominant emission from the
smolder-ing phase of wildfires.
For the SAs and cluster-dilute aggregates observed at MCOH
andduring CARES, we calculated their mobility diameters Dm14,38
andmass distributions (Fig. 4 and 5) based on their single-particle
pro-jected area equivalent diameters14. The SAs had a range of
Dmbetween 1 and 20 mm with a mean Dm < 3 mm. The monomernumber
size distribution could be described by a mono-modal log-normal
size distribution with a mean monomer diameter of 50 nmand a
standard deviation of 5 nm. We observed a majority (95%) ofSAs in
the third stage, Da , 0.3 mm, of the impactor used for
aerosolcollection. Although characterized by very large geometric
diameters(Dm), the low Da of SAs suggest that they are highly
porous, have loweffective densities, and could get deposited in the
innermost lungairways and alveoli via the process of diffusion
deposition, similarto soot aggregates10,39. However, the extent of
lung penetration of SAsdepends on particle Da40. Super-micron size
porous carbon aggre-gates, similar to soot SAs in this study, have
been synthesized in thelaboratory41, and were observed to have
effective particle densities aslow as 2.5 mg/cm3.
DiscussionSuperaggregate formation mechanism. SAs are formed
whencluster-dilute aggregates enter into a cluster-dense
aggregationregime in flames35. This regime, defined by a small
ratio of themean aggregate nearest-neighbor separation to aggregate
size andby enhanced kinetics, results in the aggregates sticking
together andpercolating to form a volume spanning SA with a
universal Df