Chemical components of organic aerosol (OA) selectively absorb light at short
wavelengths. In this study, the prevalence, sources, and optical importance
of this so-called brown carbon (BrC) aerosol component are investigated
throughout the North American continental tropospheric column during a
summer of extensive biomass burning. Spectrophotometric absorption
measurements on extracts of bulk aerosol samples collected from an aircraft
over the central USA were analyzed to directly quantify BrC abundance. BrC
was found to be prevalent throughout the 1 to 12 km altitude measurement
range, with dramatic enhancements in biomass-burning plumes. BrC to black carbon (BC) ratios, under background tropospheric conditions, increased with
altitude, consistent with a corresponding increase in the absorption Ångström exponent (AAE) determined from a three-wavelength particle soot absorption photometer (PSAP). The sum of inferred BC absorption and measured
BrC absorption at 365 nm was within 3 % of the measured PSAP absorption
for background conditions and 22 % for biomass burning. A radiative
transfer model showed that BrC absorption reduced top-of-atmosphere (TOA) aerosol
forcing by
Carbonaceous components of atmospheric aerosols are known to affect climate through direct scattering and absorption of solar radiation. The most prevalent carbonaceous aerosol component is the organic aerosol (OA) fraction, which until recently was assumed to only scatter light and act to cool the climate (Koch et al., 2007; Myhre et al., 2008). The black carbon (BC) fraction efficiently absorbs light and substantially warms the atmosphere (Bond et al., 2013). Globally, BC forcing is ranked the third most important anthropogenic climate-warming agent after carbon dioxide and methane (IPCC, 2013), considering both direct and indirect effects. Recent studies have shown, however, that components of OA also contribute to light absorption (e.g., Yang, et al., 2009; Zhang et al., 2011, 2013) and that their influence on climate may be substantial (Bahadur et al., 2012; Chung et al., 2012; Feng et al., 2013; Park et al., 2010). These compounds are referred to as brown carbon (BrC hereafter) because they tend to absorb light most efficiently at short wavelengths.
A variety of studies have investigated the sources for BrC. Primary BrC is
known to be emitted directly from incomplete combustion of fossil or biomass
fuels (Hoffer et al., 2006; Lukacs et al., 2007; Andreae and Gelencsér, 2006), and secondary organic aerosol (SOA) formed
from combustion emissions may also be brown (Saleh et al., 2013; Zhang et al., 2013). Laboratory
studies find that light-absorbing secondary compounds (chromophores) can be
formed by a variety of mechanisms, including aromatic-SOA production under
high levels of nitrogen oxides (NO
BrC can be difficult to identify with aerosol optical instruments. Lack and Langridge (2013) suggested that the use of an observed wavelength dependence of light absorption, described by the absorption Ångström exponents (AAE), to predict BrC absorption leads to substantial uncertainties. Difficulties arise because optical instruments cannot measure BrC independently of BC. Typically, BrC is determined based on differences between the observed absorption at low wavelengths, where BrC absorption is effective, to what is expected from BC alone. Both factors in the difference calculation are uncertain. There have also been attempts to decouple the effects of BC absorption (including enhancement due to internal mixing – lensing) and BrC absorption by integrating measurements with Mie theory calculations (e.g., Lack et al., 2012; Saleh et al., 2014). However, one of the main uncertainties is related to what BC absorption should be, independent of other absorbers. Some studies indicate that BC mixing state with non-absorbing materials can lead to substantial shifts in AAE relative to pure BC (Lack and Langridge, 2013), making it difficult to attribute enhanced light absorption at low wavelengths to BC mixing state or BrC, if based solely on AAE.
A more definitive and sensitive approach for identifying BrC is possible by
directly measuring chromophores in aerosol solution extracts, since the
method can isolate BrC from other absorbers (BC and mineral dust) and
long-path absorption cells provide a measurement with high sensitivity.
Studies have shown that
Chromophores in the ambient aerosol that produce the observed BrC optical
properties are not well characterized. Zhang et al. (2013) identified a
number of water-soluble nitro-aromatic compounds responsible for BrC in Los
Angeles SOA, but they only accounted for
Although studies of BrC based on aerosol extracts have been used to
investigate the sources, extent, and chemistry of fine particulate BrC, it is
difficult to use this method to assess optical properties of BrC-containing
particles. To estimate optical properties from solution data, Liu et al. (2013) used
size-resolved measurements of aerosol extract light absorbance from several
sites to estimate light absorption (
Filters were collected from the NASA DC-8 research aircraft, based out of Salina, KS, between May and June 2012 as part of the Deep Convective Clouds and Chemistry (DC3) campaign (Barth et al., 2014). The study area focused on the central USA and filter samples were obtained from near-surface to an altitude of roughly 12 km a.s.l. (pressure altitude). Figure 1a shows the locations of filter sampling periods during the study, color-coded by altitude.
The filter sampling system captured particles nominally smaller than 4.1
Filters were extracted first in 15 mL of high-purity water (18.3 M
Nomenclature.
Following water extraction, the extraction vial and filter were drained and
dried by inverting; then the filter was re-extracted in 15 mL of methanol
(VWR International, A.C.S. Grade) following the same procedure as the water
extract; however, in this case only the UV–vis absorption spectra were
measured. The estimated LOD for methanol-soluble light absorption at a
wavelength of 365 nm (MeOH_Abs(365)) is 0.11 M m
A number of gases were used in this study as emissions tracers. Biomass burning was identified using acetonitrile (CH
Particle light absorption coefficients (
Refractory black carbon (rBC; here referred to just as BC to minimize
confusion with BrC) mass concentrations were measured with a SP2 (Single
Particle Soot Photometer) and corrected to account for accumulation-mode BC
at sizes outside the detection range of the instrument (Schwarz et al., 2008). The
instrument was calibrated with fullerene soot (Alfa Aesar Lot #F12S011),
the accepted calibration material for the instrument (Baumgardner et al., 2012). Estimated
uncertainty is 30 % from flow and BC mass calibrations and aspiration
efficiency. OA was measured with a high-resolution time-of-flight Aerodyne
aerosol mass spectrometer (AMS) (DeCarlo et al., 2006). The AMS was operated with a setup
similar to that described in Dunlea et al. (2009) and using a
pressure-controlled inlet (Bahreini et al., 2008). The AMS collection
efficiency was estimated using the composition-dependent formulation of
Middlebrook et al. (2011) as implemented in the standard AMS data analysis
software (Sueper,
2015),
and applied with a 1 min time resolution
to reduce the effect of high-frequency noise. The AMS uncertainty for OA (2
The ambient light scattering coefficients (
Excluding biomass-burning plumes, the study mean
In the following analysis, we first use data on light absorption of the aerosol extracts to investigate sources and distributions of BrC. Following this, the solution data are converted to estimates of BrC aerosol absorption coefficients and the optical effects of BrC are investigated.
During the DC3 campaign, 2 out of 22 aircraft flights were specifically
targeted to investigate biomass-burning emissions, and in 6 other flights
at least one biomass-burning plume was encountered. For this work, the data
are simply delineated between clearly evident biomass-burning sampling
periods and all else, the latter being referred to as background
tropospheric conditions. To identify biomass-burning plumes, CO and
CH
Periods of identified biomass-burning plumes.
A summary of the DC3 BrC solution measurements, together with other relevant species, is given in Table 3. BrC in filter extracts was observed throughout the study region, with over 85 % of the data above the LOD.
Statistical summary of observed species throughout all flights during DC3 separated into three categories: all samples, samples during identified biomass-burning events and samples for background conditions (periods when data could not be clearly identified as biomass burning). For statistical purposes, one-half the LOD value is substituted for data below LOD. All data have been merged to the nominally 5 min filter sampling time.
Comparing with background conditions, biomass-burning plumes were notable by
significantly higher BrC levels. Average H
The proportion of water- to methanol-soluble BrC was different in background
versus biomass-burning plumes. In background air masses, the water-soluble
BrC fraction was roughly 25 to 33 % (i.e., H
Associations between various species in both the biomass-burning plumes and
under background conditions are investigated based on correlations. Results
are summarized in Table 4. Correlations are shown for
H
For biomass-burning samples, species expected from the smoke plumes (e.g.,
CO, acetonitrile, OA, WSOC, BC, and PSAP
Pearson rank correlations (
In contrast, for background conditions there was a poor correlation between
H
Total_Abs(365) was not well correlated with any of the other parameters in the background samples. This lack of correlation suggests that much of the background tropospheric BrC had undergone some form of processing or evolution (e.g., photobleaching). A similar situation is observed for WSOC, which was also not generally correlated with any of the other variables in Table 4. In a ground-based study, Liu et al. (2013) observed higher relative levels of water to methanol-soluble BrC near sources compared to aged aerosol, consistent with an aging process that preferentially depletes water-soluble fraction of BrC. Other chemical processes are possible; e.g., Lin et al. (2014) showed that IEPOX-derived oligomers that absorb in the ultraviolet region are more soluble in methanol, suggesting the potential contribution of biogenic SOA especially in biogenic-rich regions. However, a recent study showed that at a remote surface site in the southeast USA, significantly impacted by biogenic SOA, biomass-burning was the dominant BrC source (Washenfelder et al., 2015). Therefore, we believe aged biomass burning is the main source of the ubiquitous BrC, but that biogenic SOA cannot be ruled out.
These data show that BrC is detected throughout the continental troposphere.
Vertical profiles of H
Vertical profiles of absorption measured in filter water extracts and the sum of water and methanol extract (total), both at 365 nm, SP2 BC concentration, and PSAP absorption at 660 nm. Data are binned into 1km ranges and the median values are shown. Error bars indicate inter-quartile ranges. The column in the middle shows the number of data points in each altitude bin, with black for background conditions (upper) and red for biomass burning (bottom row).
BC concentration and PSAP
Vertical profile of the relative cumulative fraction (summed over all altitudes above vs. the total column), for BC (SP2), BrC at 365 nm based on extract solution absorption, PSAP absorption at 660 nm, and estimated PSAP total aerosol absorption at 365 nm, during background tropospheric conditions.
To compare vertical distributions of aerosol light-absorbing components in
background air masses, the cumulative column fraction of light absorption
coefficient or BC concentration is plotted in Fig. 3. At a given altitude,
the cumulative fraction is the light absorption coefficient, or
concentration, integrated over all altitudes above, relative to the total
column. Note that the integral of the actual absorption occurring would
depend on the vertical profile of actinic flux; this is independent of the
relative distributions that we explore here. Half the column BC
concentration occurs at approximately 3.5 km a.s.l., while for water or
total (water plus methanol extract) BrC, this occurs between 5 and 6 km,
indicating a more uniform vertical distribution and not as dominated by
surface emissions compared to BC. PSAP-determined aerosol absorption at 660 nm and light absorption extrapolated to 365 nm is also shown (the method for
extrapolating is discussed below; see Eq. 7). For PSAP absorption efficiency
at 365 nm, the 50 % altitude is 4.5 km a.s.l., which is between BC and
BrC, suggesting contributions to light absorption by a mixture of BC and BrC
at 365 nm, while the 50 % altitude for PSAP absorption efficiency at 660 nm
is
The ratio of BrC (Total_Abs(365)) to the SP2-measured BC mass also increases with altitude (see Fig. S1 in the Supplement), further demonstrating the differences in vertical distributions of BrC and BC for the background troposphere. These results suggest there is in situ BrC production or possibly preferential loss of BC with increasing altitude. Higher BC at the surface may reflect greater contributions from fossil fuel combustion sources for BC. Vertical profiles of aerosols greatly affect overall radiative forcing.
The wavelength (
Example solution spectra of H
The BrC absorption Ångstöm exponents were somewhat similar for
background conditions and biomass-burning samples; however, there were
significant differences between water and methanol extracts. For water
extracts the mean
The AAE for the overall light-absorbing ambient aerosol can also be
calculated from the more limited spectral data (three wavelengths)
associated with the PSAP. Here, AAE
AAE altitude profiles are plotted in Fig. 5. On average, there is no
significant variability in the vertical profiles of background air-mass mean
AAE
Vertical profiles of absorption Ångström exponent (AAE) of
AAE
In the previous analysis, BrC solution light absorption data were presented. Now, BrC solution data are converted to optical absorption to quantify the separate contributions of BrC and BC as a function of altitude. The sum of BC and BrC absorption are then compared to the PSAP data (total BrC and BC). The analysis could be done at any wavelength; however, 365 nm is chosen since it is in a wavelength range where a reliable BrC measurement is possible; e.g., at lower wavelengths, other non-BrC species begin to impact the data, such as nitrate, but sufficiently low that BrC, if present, should have a significant optical effect (i.e., BrC absorption drops off rapidly with increasing wavelength, as seen above, Fig. 4).
To convert the solution absorbance to light absorption by an aerosol,
knowledge of both particle morphology and how the chromophores are
distributed amongst particle size is needed. In the past, studies have often
assumed a small particle limit when making this conversion, where light
absorption by BrC aerosol is taken as 0.69 to 0.75 times the light
absorption of the solution (e.g., Sun et al. 2007). This likely gives a lower limit
for BrC absorption since BrC is not associated with sub-nanometer size particles.
Liu et al. (2013) measured the size distribution of BrC and showed that the
chromophores were consistently found in the accumulation mode in both fresh
vehicle emissions and for more aged background aerosols (BrC geometric mass
mean diameter was
From Mie theory calculations, assuming that the BrC was externally mixed
with other absorbers, Liu et al. (2013) found that aerosol absorption is approximately
1.8 to 2 times higher than the bulk absorption measured in the extracts.
Washenfelder et al. (2015) used bulk measurements of BrC absorption at 365 nm to estimate the OA
refractive index and used OA size distributions and Mie theory and also
found a conversion factor of 2 for aerosols at a rural site. In this
study, by applying Mie theory to AMS-measured size-resolved OA assuming that
the BrC is evenly distributed amongst all OA (details described in
Washenfelder et al., 2015), we obtain a conversion factor of 2.08
Considering the known uncertainty in the conversion factor of 2 (estimated to be at least 30 %; Liu et al., 2013) and the liquid absorption measurements, the overall uncertainty of these coefficients is estimated to be at least 30 and 45 %.
To estimate light absorption by the ambient aerosol at 365 nm, PSAP
measurements at higher wavelengths are extrapolated to 365 nm using a
calculated AAE
It is noted that AAE
Since data on light absorption by BC are not available, it was estimated. A
number of possible methods are available. In the first case, BC absorption
at a certain wavelength (
Aged BC aerosol is likely to be internally mixed with other aerosol
components, which, based on simplified models, such as spherical clear
shells over absorbing BC cores (Bond et al., 2006), and limited laboratory data
(Schnaiter et al., 2005; Slowik et al., 2007), could lead to a significantly different AAE
Instruments that measure light absorption based on particles deposited on a
filter, such as the PSAP, can also be significantly biased high due to
artifacts (Lack et al., 2008). To avoid this, an alternative approach to calculate BC
absorption is to estimate the light absorption coefficient at high
wavelengths, where BrC does not absorb light, using the BC mass
concentration and an assumed characteristic BC mass absorption cross section
(MAC) at a given wavelength. BC absorption at other wavelengths can be
determined using the AAE
For consistency, this prediction of BC absorption is compared to the ambient
aerosol absorption (
A schematic showing the various optical calculations is given in Fig. 6.
Schematic of how ambient aerosol and BC absorption was extrapolated
to lower wavelengths. Square data points represent PSAP measurement, which
are used to estimate the ambient aerosol AAE (AAE
In this data set, the second approach leads to a lower prediction of BC
absorption compared to the first method (i.e.,
Vertical profiles of estimated aerosol optical absorption at 365 nm
by BrC, BC, determined by an extrapolation from PSAP absorption at 660 nm
(
The schematic in Fig. 6 suggests that light absorption by the ambient
aerosol at 365 nm (e.g.,
In the following we focus on BC absorption based on the PSAP measurements
and an assumed AAE
Vertical profiles of altitude-binned median data of the light absorption
coefficients at 365 nm for BC (
Closure analysis of
A more quantitative assessment of closure for background conditions can be
seen in a scatter plot with orthogonal distance regression of the sum of the
estimated BC and BrC versus PSAP absorption (Fig. 8). From Fig. 8a, on
average for background tropospheric conditions, at 365 nm BC accounts for
roughly 74 % of the ambient absorption. When the water-soluble BrC is
added, a slope of 0.90 (Fig. 8b) indicates that the BC plus water-soluble
BrC improves the closure, but still slightly underpredicts the light
absorption coefficient. When the total BrC is used (water
Light absorption closure was carried out based on the assumption that BrC is
externally mixed with BC. Mie theory calculations were performed using the
internally mixed assumption, with core-shell sizes estimated from OA : BC
ratios obtained from AMS and SP2 measurements. BC core refractive index was
set at 1.950–0.79
Based on these assumptions, the light absorption estimated for the core-shell is 3.3 times that of only the BC core, and will be 2.4 times of the aerosol light absorption measured by the PSAP. In contrast, assuming BrC and BC are externally mixed (no core-shell enhancements), estimated light absorption at 365 nm from the PSAP was within roughly 25 % of that assuming external mixtures (see Figs. 8 and 9). In this case, we believe the external mixing assumption provides a more reasonable closure on light absorption.
Closure analysis of
Finally, in the previous sections we showed the prevalence of BrC increases
relative to BC with increasing altitude, based on solution data. Now that
the closure analysis provides some support for the BrC absorption
coefficients at 365 nm, the fractional contribution of BrC
absorption (both water and total BrC) relative to ambient absorption can be
assessed as a function of altitude. Figure 7c shows that the fraction of BrC
substantially increases with increasing altitude, with absorption due to
total BrC (at 365 nm) accounting for
Applying a vertical distribution analysis is not possible for biomass-burning plumes, since there were limited data points for some of the altitudes, but closure analysis based on the combined data is shown in Fig. 9. The conversion factor from solution BrC to ambient aerosol absorption has not been studied for biomass-burning events (Liu et al., 2013 did not include a size-resolved measure of biomass-burning BrC), leading to some uncertainty in this closure analysis. As before a multiplication factor of 2 is used as the base case, recognizing there is uncertainty in this assumption.
Biomass burning is known to be a strong source for BrC, and Fig. 9a shows
that on average, at 365 nm BC only accounted for roughly 57 % of the light
absorption, substantially lower than that for background conditions. For BC
plus water-soluble BrC the slope is 77 % and, for BC plus total BrC the
slope is 122 %, in this case overpredicting the observed values. An
AAE
A similar analysis, but where BC and ambient light absorption are based on
an assumed BC MAC, SP2-measured BC concentrations, and PSAP AAE (i.e., see
Fig. 6;
The Santa Barbara DISORT Atmospheric Radiative Transfer (SBDART) model was
used to assess the role of BrC in direct radiative forcing for background
conditions in the continental troposphere. Vertically resolved optical
properties were used, including light absorption coefficients for BC, BrC,
and total absorption based on the PSAP, along with measurements of the light
scattering coefficients from the multi-wavelength nephelometer. Absorption
and scattering coefficients were calculated for 10 wavelengths, over the
300–700 nm range, and average values were determined for each 1 km altitude
bin. BC absorption was determined using Eq. (8) and AAE
To assess the influence of BrC relative to BC, forcing was calculated based
on the estimates of BC optical properties (AAE
Alternatively, we could use the PSAP data to estimate total light absorption
by aerosols instead of BC
Most measurements of BrC and other aerosol optical properties are made at
the surface. To allow for estimates of TOA forcing due to contributions of
aerosol BrC throughout the column, the average distributions of SSA and optical depth observed under background conditions
are
used to generate a chart relating aerosol radiative forcing efficiencies
(RF
BrC radiative forcing efficiencies, defined as the BrC TOA direct radiative forcing divided by AOD at 500 nm, as a function of BrC to BC absorption ratio and SSA measured at surface at 365 nm. The circle corresponds to average background conditions determined from the DC3 campaign. The star represents a surface measurement from southeast USA, where altitude-resolved data were not available.
If the vertical profiles applied in SBDART represents typical background tropospheric conditions in the continental USA, application of Fig. 10 is not limited to this campaign. Further airborne studies similar to this one are needed to assess this assumption. As noted, the SSA and BrC/BC absorption ratios plotted in the figure are surface values at 365 nm, while column AOD could be easily retrieved from remote sensing techniques; for example, AOD at 500 nm is available from AERONET. Therefore, the figure can serve as a look-up chart to estimate radiative forcing contributions by BrC, when altitude-resolved parameters are not available. For example, a data point for surface measurements at a rural site in the southeastern USA (Washenfelder et al., 2015) is also shown in Fig. 10. In addition, large-scale models require substantial number of computations, the patterns shown in this figure could be considered as a simplified module and incorporated into models with minimal computational costs.
Direct measurements of BrC were made on solvent extracts from filters collected at altitudes ranging from approximately 1 to 12 km over the central USA during summer. The data were segregated into periods of sampling in biomass-burning plumes and more typical background tropospheric conditions. The filters were extracted sequentially: first in water, then in methanol, and the sum of the water plus methanol extract BrC assumed to represent the total BrC.
During biomass-burning periods, both water- and methanol-soluble BrC were highly correlated with other known emissions from biomass-burning plumes, including CO, acetonitrile, and BC. Under background conditions, the water-soluble fraction of BrC was somewhat correlated with smoke tracers, whereas the methanol-soluble BrC was not well correlated with any specific tracers, but most correlated with WSOC, possibly due to the BrC evolving to a more water-insoluble state as it aged. BrC was 4 to 5 times higher in biomass-burning plumes relative to the background conditions and more water-soluble (45 % of the total, at 365 nm, in the biomass-burning plumes versus 30 % in background air).
BrC was found throughout the tropospheric column. For background conditions, BrC was more evenly distributed throughout the column than BC, resulting in an increasing proportion of BrC relative to BC, with increasing altitude. This was consistent with an observed increasing AAE from the three-wavelength PSAP data.
Estimates of BC and BrC absorption coefficients at 365 nm were compared to
observed PSAP absorption. For background air masses, a closure between BC
To estimate the BrC contribution to climate forcing, the vertically resolved data were applied to a radiative transfer model (SBDART). The overall negative TOA forcing by aerosol scattering was reduced by approximately 20 % due to BrC absorption. Thus, for the aerosol loadings recorded during this study under background conditions, BrC increased the shortwave solar absorption in the atmosphere by roughly 20 % over what would occur if it were not present. These results demonstrate that BrC is an important climate forcing agent and should be considered in global climate models.
Because there are differences between BC and BrC sources and vertical distributions, the latter has an impact on the radiative forcing (Samset and Myhre, 2011); BrC cannot be accurately represented by a simple scaling of BC. Furthermore, this study and others (Lee et al., 2014; Zhong and Jang, 2014) show that BrC is dynamic with significant changes possibly occurring with photochemical aging, making predicting BrC levels and optical effects of BrC absorption complex. Instead a look-up chart was developed based on the average vertical profile for atmosphere background conditions in this study. The chart provides estimates of BrC radiative forcing based on three surface-measured aerosol parameters. The look-up chart is an important first attempt at developing a tool to assess the role of BrC radiative forcing and aid in including BrC in global models. Measurements of well-aged BrC vertical profiles similar to those of this study are needed in other locations to improve the predictability of this type of model.
This project was funded by GIT NASA contracts NNX12AB83G and NNX08AH80G and UNH NASA contract NNX12AB80G. Acetonitrile measurements onboard the DC-8 were supported by BMVIT/FFG-ALR and the NASA Postdoctoral Program. P. Campuzano-Jost, D. A. Day, and J. L. Jimenez were supported by NASA NNX12AC03G. The authors thank the DC3 personnel for logistical support.Edited by: N. M. Donahue