Transport of anthropogenic aerosol into the Arctic in the spring
months has the potential to affect regional climate; however, modeling
estimates of the aerosol direct radiative effect (DRE) are sensitive to
uncertainties in the mixing state of black carbon (BC). A common approach in
previous modeling studies is to assume an entirely external mixture (all
primarily scattering species are in separate particles from BC) or internal
mixture (all primarily scattering species are mixed in the same particles as
BC). To provide constraints on the size-resolved mixing state of BC, we use
airborne single-particle soot photometer (SP2) and ultrahigh-sensitivity
aerosol spectrometer (UHSAS) measurements from the Alfred Wegener Institute
(AWI) Polar 6 flights from the NETCARE/PAMARCMIP2015 campaign to estimate
coating thickness as a function of refractory BC (rBC) core diameter and
the fraction of particles containing rBC in the springtime Canadian high
Arctic. For rBC core diameters in the range of 140 to 220 nm, we find
average coating thicknesses of approximately 45 to 40 nm, respectively,
resulting in ratios of total particle diameter to rBC core diameters ranging
from 1.6 to 1.4. For total particle diameters ranging from 175 to 730 nm,
rBC-containing particle number fractions range from 16 % to 3 %,
respectively. We combine the observed mixing-state constraints with simulated
size-resolved aerosol mass and number distributions from GEOS-Chem–TOMAS to
estimate the DRE with observed bounds on mixing state as opposed to assuming
an entirely external or internal mixture. We find that the pan-Arctic average
springtime DRE ranges from
Over the last several decades, the Arctic has warmed at nearly twice the rate
of the global mean (Boucher et al., 2013). In addition to
The population (or chemical) mixing state of BC refers to the degree to which BC particles are mixed with other aerosol species. In global and regional models in which population mixing state is not explicitly tracked, BC is commonly represented as completely externally mixed (separate from other aerosol species) or completely internally mixed (mixed together with other aerosol species). For BC mass in a given particle size range, the external mixture assumes the BC mass is divided up into particles of the same size as the non-BC particles. Conversely, in the internal mixture, the same BC mass is divided up into a relatively larger number of particles with smaller BC diameters such that the total mixed particle has the representative diameter of the size range. In a full internal mixture, BC mass is spread among all particles. For a given BC mass, the internal mixing-state assumption will produce more absorption as the BC mass is spread amongst more particles, giving rise to greater surface area and thicker coatings (Bond and Bergstrom, 2006; Seinfeld and Pandis, 2012). However, in the atmosphere, BC is often not mixed entirely as an internal or external mixture. For instance, Massoli et al. (2015) found that 35 % of the total non-refractory submicrometer mass was associated with BC-containing particles in coastal California during the CALNEX campaign, and recent studies of Arctic aerosol suggest that a smaller percentage of aerosol particles contain BC (e.g., Raatikainen et al., 2015; Sharma et al., 2017). Thus, the assumption in aerosol models of a full internal or external mixture may lead to overestimates or underestimates of BC absorption.
A second component of BC mixing state is the morphological mixing state. The morphological mixing state refers to the distribution of chemical species within a particle, as well as the shape and location of BC within the aerosol particle. At a remote observation site, China et al. (2015) found substantial variability in the fractal dimension and structure of mixed (i.e., fully encapsulated versus partly encapsulated) BC particles. Despite this variability, a common assumption for the morphological mixing state of BC is that the BC mass forms the core of a particle and the hydrophilic aerosol mass forms a concentric shell around the particle (Bond et al., 2006; Lack and Cappa, 2010). The scattering component of the shell acts as a lens to focus more photons onto the core, thus increasing absorption (Borhen and Huffman, 1983; Bond et al., 2006). However, the degree of absorption enhancement is a strong function of the structure and geometry of the mixed particle as well as the core diameter and shell thickness (He et al., 2015, 2016). Theoretical calculations (known as “core–shell Mie theory”) and laboratory studies have estimated enhancements in absorption for a given BC mass of a factor of 1.3 to 2 (Schnaiter et al., 2003, 2005; Zhang et al., 2008). Based on these findings, Bond et al. (2006) recommend scaling the BC absorption by a factor of 1.5 to account for the “lensing” effect in models that do not assume internal aerosol mixtures. However, field campaigns have found a wider variation of absorption enhancements. While finding that 20 %–30 % of BC particles had acquired a thick coating, McMeeking et al. (2011) and Subramanian et al. (2010) did not find a dependence of mass absorption cross section with coating. Similarly, Cappa et al. (2012) found an absorption enhancement of only 6 % and speculated that the BC may not be at the exact center of the particle. Conversely, Wang et al. (2014) found an absorption enhancement of 1.8 in China, similar to laboratory findings. In addition, Sharma et al. (2017) found values of mass absorption cross section to increase with coating thickness with a steeper slope than theoretical calculations. Thus, accurate model estimates of BC absorption must rely on an understanding of sources of BC and all other aerosol components.
Aerosol number and mass concentrations in the Arctic have a strong seasonal cycle, with contributions from anthropogenic sources leading to a peak in accumulation-mode aerosol mass in the winter and spring (e.g., Quinn et al., 2007; Croft et al., 2016a). Efficient wet removal in the summer results in conditions that favor new particle formation and nucleation-mode aerosol (e.g., Garrett et al., 2011; Browse et al., 2012; Leaitch et al., 2013; Tunved et al., 2013; Croft et al., 2016a, b; Willis et al., 2016). As there is little or no solar irradiance in the Arctic winter, it is essential to accurately simulate aerosol optical properties in the spring. Sources of BC in the Arctic include gas flaring and biomass burning; however, much of the atmospheric BC concentration is transported from lower latitudes (Xu et al., 2017; Qi et al., 2017a; Schulz et al., 2018). Using a combination of observations and a chemical transport model, Xu et al. (2017) found that in the Arctic spring much of the BC concentrations at higher elevations arrived from South Asia, while at lower elevations BC concentrations were transported from Asia and Eastern Europe. Similarly, Qi et al. (2017a) found BC concentrations in the Arctic in April 2008 to be largely from anthropogenic sources in Asia and biomass burning sources in Siberia. Sources of non-BC particles to the Arctic include direct marine sources, new particle formation from natural or possibly anthropogenic precursors, and transport from lower latitudes (e.g., Croft et al., 2016b; Wentworth et al., 2016; Willis et al., 2016).
Despite the dependence of aerosol absorption on BC mixing state, measurements of mixing state in the Arctic are limited. The single-particle soot photometer (SP2) provides direct measurements of size-resolved refractory BC (rBC) mass for particle diameters in the range of approximately 75–700 nm. In addition, through the leading-edge fit method (Gao et al., 2007), the SP2 can provide estimates of coating thickness on rBC particles. Combining the size-resolved coated rBC measurements with a total aerosol size distribution provides information on the fraction of total aerosol number containing rBC, thus providing a constraint on the population mixing state of rBC. Raatikainen et al. (2015) measured rBC mixing-state properties with an SP2 in the Finnish Arctic winter, finding 24 % of particles contain rBC with an average total particle to refractory rBC core diameter of 2. Similarly, Liu et al. (2015) found an average total particle to rBC core diameter of 2.25 (1.7–2.8) during an aircraft campaign in the European Arctic in March. For biomass burning plumes, Kondo et al. (2011) measured a ratio of 1.3–1.6 during aircraft flights in the spring and summer in the Canadian Arctic. Finally, Sharma et al. (2017) found a ratio of 1.4–1.25 at the Alert ground station in spring of 2012.
In this study, we present measurements of BC mixing state aboard the Alfred Wegener Institute (AWI) Polar 6 flights as part of the Network on Climate and Aerosols: Addressing Key Uncertainties in Remote Canadian Environments (NETCARE) and Polar Airborne Measurements and Arctic Regional Climate Model Simulation Project (PAMARCMiP). We use these measurements as constraints on the population mixing state to estimate the direct radiative effect (DRE) in the springtime Arctic and compare these estimates to the DRE calculated using bounding cases of completely external or internal mixing-state assumptions. Note that these measurements do not allow us to constrain the morphological mixing state. While this study focuses on the DRE, we note that other aerosol–radiation interaction processes, such as the semi-direct effect, may also depend on BC mixing state. In Sect. 2.1–2.3, we present the Polar 6 flight paths and size-resolved aerosol measurement setup. In Sect. 2.4, we discuss the chemical transport model and assumptions in the calculation of the DRE. In Sect. 3, we present observations of BC mixing state and implications for simulated DRE. We share our conclusions and study limitations in Sect. 4.
Map showing the location of Alert and Eureka as well as the flight paths for the six flights in the Canadian high Arctic portion of the Polar 6 campaign. The flights were undertaken between 7 and 13 April 2015.
As part of the Network on Climate and Aerosols: Addressing Key Uncertainties
in Remote Canadian Environments project (NETCARE,
Aerosol was sampled near-isokinetically through a stainless steel shrouded
diffuser inlet located ahead of the engines. The inlet provided near-unity
transmission of particles 20 nm to
Periods when the aircraft was in-cloud were determined using data from a
forward-scattering spectrometer cloud probe (FSSP-100; Droplet Measurement
Technologies, Inc). The FSSP is an optical particle counter that detects
droplets in the range of 2–50
Measurements of refractory black carbon (following the definition in Petzold
et al., 2013) were made with a single-particle soot photometer (SP2; Droplet
Measurement Technologies Inc). The SP2 detects individual particles using an
intra-cavity Nd:YAG laser operating at 1064 nm. As particles pass through
the laser beam, those that contain a strongly absorbing component at 1064 nm
(such as black carbon) are heated to incandescence. Light emitted by the
incandescing fraction of the particle is detected by a pair of
photomultiplier tubes. The peak amplitude of the thermal radiation emitted by
the incandescing particle is proportional to the mass of refractory material
in the particle (Moteki and Kondo, 2007; Slowik et al., 2007). In this work,
size-selected Aquadag particles (Acheson Industries) were used as an external
standard for mass calibration of the SP2. Measured Aquadag mobility diameters
were converted to rBC (refractory black carbon) mass using the size-dependent
effective densities reported by Gysel et al. (2011). Recent studies have
shown that the SP2 is more sensitive to Aquadag than it is to ambient rBC
(Laborde et al., 2012; Moteki and Kondo, 2010). In order to account for
this, we have scaled the slopes of the Aquadag-derived calibration curves by
a factor of
Two single-particle soot
photometers (referred to as SP2-1 and SP2-2) were used in this study. SP2-1
had a detection range of 0.40–323 fg rBC (equivalent to spherical
diameters ranging from 75 to 700 nm at an rBC density of 1.8 g cm
As particles pass through the laser beam in the SP2 they elastically scatter
light at 1064 nm. For particles that contain rBC, the particle may be heated
and begin to vaporize before the scattering intensity reaches the peak value
that an unperturbed particle would have reached. In this case, the
unperturbed scattering amplitude can be retrieved by fitting the leading edge
of the particle's scattering signal as described by Gao et al. (2007). In the
leading-edge fit method, the center position and width of a Gaussian (which
reflects the laser beam profile) are fixed from the scattering profiles of
non-incandescing particles collected during the preceding and following hour.
The unperturbed peak scattering amplitude for an incandescing particle is
then determined by fitting the measured scattering profile up to 5 % of
the peak scattering intensity using the fixed width and center position and
allowing only the amplitude to vary. Fitting the profile only up to 5 %
of peak elastic scattering intensity allows the unperturbed peak amplitude to
be retrieved. For an individual particle the uncertainty in reproducing the
unperturbed amplitude is
With the scattering amplitude determined by the leading-edge technique and
the measured rBC core diameter, a core–shell Mie model can be used to
determine the optical diameter of the rBC-containing particles. In the Mie
model we used a refractive index of 2.26–
Although the SP2 can measure individual particle rBC mass down to 0.40 fg
(
Total aerosol size distributions for particles with diameters in the range of
85–1000 nm were measured by an ultrahigh-sensitivity aerosol spectrometer
(UHSAS; Droplet Measurement Technology Inc). The UHSAS is a laser-based
aerosol spectrometer in which particles intercept the beam of a solid-state
Nd3
Schematic presenting the different BC mixing states for the TOMAS
size bin corresponding to particle diameters of 250 nm. The bold text shows
the parameter being constrained in each mixing state. In this example,
GEOS-Chem–TOMAS simulates 4 % BC mass fraction and a particle number
concentration of 100 cm
One potential issue with using the UHSAS aboard an aircraft is that as the
pressure changes during ascent and descent, the sample flow at the inlet of
the chamber can deviate from the measured and regulated flow at the outlet of
the chamber (Brock et al., 2011; Kupc et al., 2018), which results in
inaccurate particle concentration measurements. Brock et al.
(2011) saw particle number deviations from a reference counter of
To determine the fraction of total aerosol particles containing rBC as a function of size, we first determined the number size distribution of the rBC-containing particles, this time accounting for both the rBC core diameter and the thickness of any coating material. This was done by taking the number size distribution of the rBC cores determined from SP2-1 (Sect. 2.2.1) and applying the coating thicknesses as a function of rBC core size determined from SP2-2 (Sect. 2.2.2). Once the number size distribution of the rBC-containing particles was determined, the fraction of rBC-containing particles as a function of size was determined by dividing the number size distribution of the rBC-containing particles by the number size distribution of the total aerosol determined by the UHSA (Sect. 2.3). As with the coating analysis, this process was carried out separately for each flight and the averages of all flights were combined with GEOS-Chem–TOMAS simulations. Additional details are given in the Supplement.
To simulate aerosol concentrations in the Arctic, we use the Goddard Earth
Observing System chemical transport model, GEOS-Chem version 10.01. We
simulate April 2015 with 2 months of spin-up not included in the analysis.
Transport in GEOS-Chem is driven by MERRA reanalysis meteorology fields. This
version of GEOS-Chem uses a horizontal resolution of 4
Aerosol microphysics is simulated using the TwO-Moment Aerosol Sectional
(TOMAS) microphysics scheme (Adams and Seinfeld, 2002) coupled with GEOS-Chem
(known as GEOS-Chem–TOMAS). The version of TOMAS used in this study includes
40 size bins ranging from diameters of approximately 1 nm to 10
Global anthropogenic emissions are derived from the Emissions Database for Global Atmospheric Research (EDGAR) Hemispheric Transport of Air Pollution (HTAP) version 2.2 (Janssens-Maenhout et al., 2015). Following the recommendation in Xu et al. (2017), we include BC and organic carbon emissions from gas flaring derived from the Evaluating the Climate and Air Quality Impacts of Short-Lived Pollutants (ECLIPSE) emission inventory (Klimont et al., 2017). Biomass burning emissions are from the Fire Inventory from NCAR (FINN) for the year 2015 (Wiedinmyer et al., 2011). Dust aerosol emissions follow the DEAD scheme (Zender et al., 2003), while sea salt aerosol emissions are based on the scheme of Jaeglé et al. (2011).
Description of BC mixing states.
The all-sky direct radiative effect is estimated using an offline version of the Rapid Radiative Transfer Model for GCMs (RRTMG; Iacono et al., 2008), following the online version implemented in GEOS-Chem (Heald et al., 2014). RRTMG treats clouds using the Monte Carlo independent column approximation (McICA; Pincus et al., 2003). Aerosol optical properties are calculated using monthly averaged aerosol mass and number concentrations with refractive indices from the Global Aerosol Dataset (GADS). We use Mie code published in Bohren and Huffman (1983) to calculate aerosol optical depth, single scattering albedo, and asymmetry parameter. Available code includes Mie calculations for two concentric spheres (for use in core–shell morphologies). The use of monthly mean aerosol and cloud properties is a limitation of this study; however, we feel this is sufficient to explore the impacts of different BC mixing-state assumptions on the DRE.
We calculate the DRE with five BC mixing states, which are outlined in
Table 1. Figure 2 is a schematic of the mixing states. We discuss a numerical
example at the end of this section that follows Fig. 2. In the
We compare these assumed mixing states to two mixing states based on the
measurements described in Sect. 2.2–2.3. These measurements constrain only
the population mixing state. For cases of mixed BC, we assume an ideal
core–shell mixture, but note that this morphological mixing state is an important
uncertainty (China et al., 2015). The
The second measurement-constrained mixing state,
Coating thickness as a function of rBC core diameter. Grey markers are the median, dark shaded area is the 25th–75th percentile, lighter shaded area is the 10th–90th percentile of coating thicknesses for each bin.
The different mixing states are depicted schematically in Fig. 2 with the
bold text highlighting the parameter being constrained in each case. As an
example, we depict the TOMAS size bin for 250 nm diameter particles for
which
GEOS-Chem–TOMAS simulates a particle number concentration of 100 cm
Figure 3 shows measured coating thickness as a function of the volume-equivalent diameter (VED) of the rBC cores (both the rBC cores and the coating are assumed to be spherical). The alternate axis gives the fraction of detectable notch positions (see Sect. 2.2.4 for details). The black dots represent the median and the shaded regions represent the interquartile range and 10th–90th percentile range of measurements for all particles across the different flights and altitudes. For rBC core diameters ranging from 140–220 nm (the region with greater than 90 % successful fits), the median measured coating thickness decreases slightly from 45 to 40 nm (with an interquartile range of 30–70 to 17–65 nm). This results in total particle to rBC core diameter ratios ranging from 1.6 (IQR: 1.4–2.0) for rBC cores at 140 nm to 1.4 (IQR: 1.2–1.6) for rBC cores at 220 nm. This range is similar to measurements in the Canadian Arctic by Kondo et al. (2011). When combining with model results, we use only the measured core and shell thicknesses in the range with greater than 90 % detectable notch positions (i.e., core diameters larger than 140 nm) and use minimum and maximum bounding assumptions in the region 70–140 nm. Figure S2 shows the minimum and maximum shell thickness across the range 70–700 nm. A detailed examination of vertical variability in rBC measurements from this campaign is included in Schulz et al. (2018).
Number distributions for the bare rBC cores (black), the coated
rBC-containing particles (green), and the total aerosol as detected by the
UHSAS (grey) for the median coating thickness scenario. Shown in the bottom
panel is the distribution of the fraction of the total aerosol containing rBC
in the median coating scenario. The solid black line is a polynomial fit to
the observations following the form (with
Across all flights (and altitudes), we do not find substantial variability in measured coating thickness. This lack of variability is likely due to atmospheric processing over several days during transport of the air mass to the Arctic region. In the Supplement, we plot coating thickness for each flight in Fig. S3. The low degree of variability across flights justifies the use of campaign averages in our model analysis.
Figure 4a shows the measured number distributions for uncoated rBC, coated rBC, and total particle number, with sets of symbols representing the average across an individual flight. The diameters of the coated rBC particles represent a sum of the rBC core diameter and twice the shell thickness. The total aerosol size distribution has a mode centered near 150 nm, similar to measured accumulation-mode particles in the Arctic spring reported in Croft et al. (2016a). The bare rBC size distribution peaks below the 70 nm size limit of the SP2. To estimate the size-resolved fraction of particles containing rBC, we take the ratio of the SP2-coated rBC particles and the UHSAS total particle size distribution (Fig. 4b). The solid line in Fig. 4b is the best fit to the experimental data shown in Fig. 4b. A polynomial function was used because this gave a good fit to the experimental data. The coefficients for this fit are provided in the caption to Fig. 4; however, we note that this fit is derived only from measurements in the springtime Arctic and is not based on any underlying understanding of the chemical or physical processes involved. With this polynomial fit, we find that 16 % of particles in the 175–330 nm diameter size range contain rBC, with values ranging from 10–29 %. For particles with diameters ranging between 550 and 730 nm, this fraction decreases to 3 %. This result is similar in magnitude to surface measurements of rBC-containing particle fractions from Sharma et al. (2017) for Alert, Canada. In general, we expect chemically aged plumes to be more internally mixed (through condensation and coagulation). The relatively low fractions of rBC-containing particles measured here may suggest local sources of non-BC accumulation-mode particles, such as sea salt, which is common in April at Alert (e.g., Leaitch et al., 2018), or slow coagulation timescales.
The fraction of BC aerosol mass relative to total aerosol mass at each size bin in GEOS-Chem–TOMAS at three vertical levels along with the fraction determined from the SP2 and UHSAS observations.
Figure 5 shows the size-resolved fraction of BC mass relative to total
aerosol mass based on the ratio of the SP2 to USHAS compared to
GEOS-Chem–TOMAS simulations. Due to instrument constraints, the measured size
range is restricted to diameters of 175–730 nm compared to the TOMAS size
range of 1 nm to 10
The simulated pan-Arctic mean shell thickness as a function of BC
core diameter when the number fraction of BC is constrained by observations
As the ratio of BC mass to total aerosol mass in TOMAS is lower than in
measurements, the two different mixing-state constraints
(
The net direct radiative effect (DRE) calculated using each mixing state averaged over the Arctic for April.
To explore the impact of mixing state on DRE in the Arctic spring, we
calculate the DRE due to all aerosol with the bounding mixing-state
assumptions and the measurement constraints on mixing state. Figure 7 shows
the net DRE over the Arctic for April assuming all BC is externally mixed.
The pan-Arctic average DRE is
The net direct radiative effect over the Arctic in April assuming externally mixed particles.
Table 2 presents the April pan-Arctic mean DRE from all aerosol for the five
different mixing-state assumptions. As expected, the DRE calculated with the
external mixing-state assumption produces the most negative DRE
(
The difference in DRE between fBC-constrained
The difference in the DRE between the measurement-constrained mixing states
(fBC-constrained and
Table S2 presents the pan-Arctic mean DRE for the
We acknowledge several limitations in understanding the difference in DRE when using measurement constraints (instrument limitations are discussed in Sect. 2). First, the measurements described here constrain only population mixing state. With regard to particle morphology, we assume a core–shell configuration with BC at the exact center of the particle. Several studies have suggested this may not always be representative of atmospheric aerosol (e.g., Cappa et al., 2012; China et al., 2015). We note that as we assume BC is always at the center of a mixed particle, BC absorption in this work is therefore an upper bound. Second, we assume all BC particles are coated (though sometimes with very thin coats as in the minimum coating assumption; see the Supplement). Some studies show that plumes may have a combination of coated and uncoated BC, with the coated BC fraction increasing with the chemical age of the plume (e.g., Subramanian et al., 2010). Third, we use averages of measurements across flights and altitudes for BC size distribution and coating information. A separate paper, Schulz et al. (2018), will examine the spatial and vertical variability of BC measurements made in this campaign. Fourth, our measurements of BC mixing state only apply to BC particles in a limited size range. The size distributions in Fig. 4 imply a substantial fraction of BC and non-BC number concentration outside the size range of our measurements. We attempt to account for this limitation in our model analysis through consideration of upper and lower bounds on coating and BC-containing particles. Finally, the measurements used here are limited spatially and temporally. Future work may expand on the measurements reported here for other months and a larger spatial domain.
In this study, we present measurements of BC mixing state in the springtime
Canadian high Arctic. Using an SP2 aboard the Polar 6 flights, we find that on
average for rBC core diameters in the range of 140–220 nm, median coating
thickness decreases from 45 to 40 nm (with an interquartile range of
30–70 nm to 17–65 nm). The ratio of total particle to rBC core diameter
in this study (1.6–1.4) is comparable to measurements in the springtime
Canadian Arctic by Kondo et al. (2011) and Sharma et al. (2017), as well as
measurements in the European Arctic from Raatikainen et al. (2015) and Liu et
al. (2015). Combining the SP2 size-resolved rBC measurements with total
aerosol size distributions from the UHSAS instrument, we estimate
approximately 16 % of particles contain rBC in the 175–330 nm diameter
range, and 3 % of particles contain rBC in the 550–730 nm diameter
range. We use these measurements separately as constraints on the BC mixing state
simulated in TOMAS. When constraining TOMAS mass and number concentrations
with the shell thickness as a function of rBC core diameter
(
We estimate the Arctic DRE in April 2015 using bounding mixing-state
assumptions of entirely externally mixed and internally mixed BC and compare
this to the DRE estimate using the measurement-constrained mixing states. The
estimated range of the DRE using the measurement-constrained mixing states
(
NETCARE (Network on Climate and Aerosols, 2015,
The supplement related to this article is available online at:
Aircraft measurements from the Polar 6 campaign as part of NETCARE were planned and carried out by SH, AB, WRL, HS, AH, MZ, JB, MW, and JA. The NETCARE project is supervised by JA. SH led the analysis of measurements of black carbon mixing state with input from AB and WRL. JP, WRL, and JK designed the use of SP2 and UHSAS measurements to constrain simulated mixing state. Model simulations and subsequent analysis was led by JK with input from JP. JK prepared the paper along with contributions from SH on the methodology of aircraft BC measurements and input from all coauthors.
The authors declare that they have no conflict of interest.
This article is part of the special issue “NETCARE (Network on Aerosols and Climate: Addressing Key Uncertainties in Remote Canadian Environments) (ACP/AMT/BG inter-journal SI)”. It is not associated with a conference.
The authors acknowledge the financial support provided for NETCARE through the Climate Change and Atmospheric Research Program at NSERC Canada, as well as support from Environment and Climate Change Canada and the Alfred Wegener Institute. This research has also been supported by a grant from the US Environmental Protection Agency's Science to Achieve Results (STAR) program through grant no. 83543801 and the US National Oceanic and Atmospheric Administration, an Office of Science, Office of Atmospheric Chemistry, Carbon Cycle, and Climate Program, under the cooperative agreement award NA17OAR430001. We also gratefully acknowledge support from SFB/TR 172 ArctiC Amplification: Climate Relevant Atmospheric and SurfaCe Processes, and Feedback Mechanisms (AC)3 in subproject C02 funded by the DFG (Deutsche Forschungsgesellschaft). Finally, this material is based upon work supported by the National Science Foundation under grant no. AGS-1559607. Edited by: Lynn M. Russell Reviewed by: three anonymous referees