Given the sensitivity of the Arctic climate to short-lived climate forcers,
long-term in situ surface measurements of aerosol parameters are useful in
gaining insight into the magnitude and variability of these climate forcings.
Seasonality of aerosol optical properties – including the aerosol
light-scattering coefficient, absorption coefficient, single-scattering
albedo, scattering Ångström exponent, and asymmetry parameter – are
presented for six monitoring sites throughout the Arctic: Alert, Canada;
Barrow, USA; Pallas, Finland; Summit, Greenland; Tiksi, Russia; and Zeppelin
Mountain, Ny-Ålesund, Svalbard, Norway. Results show annual variability
in all parameters, though the seasonality of each aerosol optical property
varies from site to site. There is a large diversity in magnitude and
variability of scattering coefficient at all sites, reflecting differences in
aerosol source, transport, and removal at different locations throughout the
Arctic. Of the Arctic sites, the highest annual mean scattering coefficient
is measured at Tiksi (12.47 Mm
The Arctic is a unique environment, characterized by sensitive interactions and feedbacks between the atmosphere, ocean, cryosphere, and biosphere (Serreze and Francis, 2006; Serreze and Barry, 2011). In recent decades, substantial changes have been observed in the Arctic, including increases in air temperature (Johannessen et al., 2004), decreases in sea ice extent and thickness (Lindsay and Zhang, 2005; Stroeve et al., 2007, 2012), changes in Arctic vegetation (Wang and Overland, 2004; Chapin et al., 2005; Pearson et al., 2013), and shifts in precipitation patterns (Groves and Francis, 2002; Bintanja and Selten, 2014). The mechanisms behind these changes are induced by anthropogenic global climate change (Anisimov et al., 2007) and have not yet been fully characterized. Human presence and thus emissions in the Arctic are likely to increase in the future due to decreases in sea ice making the region more accessible for energy extraction and shipping activities (e.g., Aliabadi et al., 2015; Eckhardt et al., 2013). More research in the Arctic, particularly on atmospheric components and processes in the region, is necessary to better understand what is changing, why it is changing, and how it might change in the future (Anisimov et al., 2007).
Within the Arctic atmosphere, short-lived climate forcers like aerosols are important contributors to the observed warming and environmental changes in the region (Quinn et al., 2008; Najafi et al., 2015). Aerosols can affect the climate both directly by scattering and absorbing incoming solar radiation and indirectly through aerosol–cloud interactions (Twomey, 1977). Quantifying the forcing by aerosols in the Arctic is especially complex, given the annual variability in surface albedo and cloudiness, the stratified atmosphere, resulting feedbacks, and long-range aerosol transport. Measurements of surface Arctic aerosol optical properties in particular can help define and constrain interannual, seasonal, and diurnal variability of light scattering and absorption, potential particle sources, and resulting radiative forcing. The observation capacity demonstrated here has potential for providing in situ observational checks on long-term black carbon inventories and monitoring strategies of importance to international pollution mitigation effects. This paper will seek to provide an overview of surface aerosol optical properties in the Arctic.
Observations of aerosols in the Arctic have a long (> 50 years) history (e.g., Mitchell's, 1957, report on so-called Arctic haze layers), although continuous surface measurements of aerosol optical properties did not begin until the mid-1970s at Barrow, Alaska (BRW), and later at other sites. The start of long-term, continuous surface measurements, ongoing to this day, have provided information about aerosol chemistry, microphysics, and optical properties and enabled the development of aerosol climatologies, the analysis of trends, and the evaluation of models. Such analyses have been driven by the need to understand the remote and local sources, transport, and processes that influence aerosol properties in the Arctic. Understanding aerosol optical properties in particular is important in gaining insight into the role of aerosols in the Arctic's radiative energy budget (e.g., Quinn et al., 2011).
Despite the challenges associated with performing high-quality, long-term
atmospheric observations in the Arctic (e.g., high costs, extreme
conditions, difficult access), several monitoring stations do
currently exist in the Arctic. Of these monitoring sites, 10 contribute to
the International Arctic Systems for Observing the Atmosphere (IASOA)
network. The purpose of the IASOA organization is twofold: (1) to enhance
interoperable observational abilities and coverage of surface atmospheric
monitoring in the data-sparse Arctic, and (2) to foster pan-Arctic
scientific collaboration with easier data access and strengthened synergy
among researchers (Uttal et al., 2016). Of the 10 monitoring sites, six
stations have multi-year, continuous measurements of aerosol optical
properties, and it is these data from 2012 to 2014 that are used for the
Arctic aerosol analysis presented in this paper. These monitoring stations
follow standardized aerosol sampling protocol, as advised by the Global
Atmosphere Watch (GAW) network
(
Published climatologies and seasonality of surface extensive aerosol optical properties (i.e., properties that depend on the amount of aerosol) have shown that, at many Arctic sites, scattering and absorption are highest in the late winter and early spring, and lowest in the summer (e.g., Bodhaine, 1983 (Barrow); Bodhaine, 1995 (Barrow); Sharma et al., 2004 (Alert); Eleftheriadis et al., 2009 (Zeppelin); Heintzenberg, 1982 (Zeppelin); Aaltonen et al., 2006 (Pallas); Lihavainen et al., 2015 (Pallas)). However, results shown here will support the notion that not all Arctic sites have this seasonal cycle. The winter/spring aerosol enhancement is called Arctic haze, referring back to Mitchell's (1957) early airborne observations. Understanding the sources, characteristics, and effects of Arctic haze has been a continuing effort over the past several decades (e.g., Rahn et al., 1977; Shaw, 1995; Quinn et al., 2007; Liu et al., 2015; and references therein). The low summertime values of absorption and scattering currently observed in the Arctic are likely to be particularly vulnerable to warmer, drier climatic conditions (e.g., due to increases in summertime forest fires and decreases in sea ice leading to enhanced marine emissions and human activities in the region during the summer). Published climatologies and seasonal cycles of in situ Arctic intensive aerosol properties (i.e., properties that are ratios of extensive properties and not directly dependent on aerosol amount) are sparse and suggest that, unlike the relatively consistent seasonal pattern for extensive properties, the seasonal cycles of intensive optical properties (e.g., Ångström exponent) may differ from site to site (Delene and Ogren, 2002; Aaltonen et al., 2006; Lihavainen et al., 2015). This work seeks to expand on previous aerosol optical analyses in the Arctic by synthesizing aerosol seasonality at multiple Arctic stations and adding new knowledge on the seasonality of intensive aerosol characteristics in the region.
At present, only surface measurements can provide a seasonal context for the range of aerosol optical properties used to determine radiative forcing efficiency (RFE), including absorption, scattering, backscattering fraction, asymmetry parameter, and single-scattering albedo. While vertical profiles are important due to the stratified conditions in the Arctic atmosphere (e.g., Rahn et al., 1977), aircraft campaigns in the Arctic thus far do not provide insight into seasonality. Stone et al. (2014; their Fig. 5) note that only one aircraft campaign in the last 30 years occurred outside the Arctic haze period. Remote-sensing instruments such as sun photometers are limited due to long periods of darkness during the winter, and satellite measurements have limited utility due to the high albedo of the Arctic snow surface and the dark Arctic winters. An additional limitation of remote-sensing measurements is that parameters important for RFE calculations (e.g., single-scattering albedo) cannot be retrieved without high uncertainties in the Arctic due to the low aerosol optical depth (AOD) (Dubovik et al., 2000). Although geographically sparse compared to the potential of remote-sensing and aircraft campaigns, surface measurements have the advantage of being long-term, year-round, and comprehensive.
The objective of this paper is to explore the seasonality and spatiotemporal variability of surface aerosol optical properties in the Arctic; the results of this exploration may be useful for continued improvement of modeling and remote-sensing capabilities. Here we ask how aerosol optical properties differ among six Arctic monitoring sites, how monthly variability in aerosol optical properties compares across the sites, what systematic variability among aerosol optical properties exists in the Arctic, what pairing of trajectory data with aerosol optical properties suggests about aerosol sources in the Arctic, and how this trajectory analysis varies geographically from station to station.
Monitoring station names, locations, scattering and absorption instruments, size cuts, and humidity of samples. Bolded instruments indicate those from which data are used in this analysis.
The analysis presented here uses in-situ-measured aerosol properties from six
Arctic monitoring stations. To be included in the analysis, a station had to
have continuous and concurrent aerosol light scattering and two sets of
absorption measurements: (i) Aethalometer and (ii) “reference” co-located
absorption instrument (details in Sect. 3.2) during
years 2012–2014. Six monitoring sites met these criteria: Alert, Canada
(ALT); Barrow, Alaska (BRW); Pallas, Finland (PAL); Summit, Greenland (SUM);
Tiksi, Russia (TIK); and Zeppelin Mountain, Ny-Ålesund, Svalbard, Norway
(ZEP) (for a record of data availability at all IASOA sites, see the IASOA
data access portal:
Map of Arctic monitoring stations with pictures of each site.
Alert is located in Nunavut, Canada, and is operated by Environment and
Climate Change Canada (ECCC). The aerosol optical property measurements are
made in collaboration with the National Oceanic and Atmospheric
Administration (NOAA). The monitoring station is the most northerly site in
the GAW network; despite the site being shared with a Canadian military
facility and an ECCC upper-air weather station, it is remote and far from
industrial pollution sources. The measurement laboratory was established in
1986 and has long-term Aethalometer measurements from 1989 on and aerosol
absorption (particle soot absorption photometer, or PSAP) and scattering measurements from 2005 on. The aerosol
instruments measure from an inlet and aerosol system that has both 1 and 10
The Barrow observatory was established in 1973 and is operated by NOAA with
additional support from the U.S. Department of Energy and the National
Science Foundation (NSF). The site is situated 5 km northeast of the town of
Barrow, Alaska (population:
The Pallas Atmosphere-Ecosystem Supersite is operated by the Finnish
Meteorological Institute (FMI) and is a part of the larger
Pallas-Sodankylä GAW station located in northern Finland. The Pallas
main research site is located in the Pallas-Yllästunturi National Park
on the top of the Sammaltunturi fell at an elevation of 565 m a.s.l. and above
the tree line. The nearest town is Muonio, located 19 km to the west with
The Summit monitoring station is located in Greenland, Denmark, and is
supported and operated by Duke University in collaboration with NOAA Earth
Systems Research Laboratory with financial aid from the NSF. The scattering
and co-located absorption measurements at SUM were initiated in 2011 as part
of a NOAA collaboration with the Georgia Institute of Technology. Summit is
unique from the other stations in this study due to its high elevation of
3238 m a.s.l., meaning it often measures free-tropospheric air. The station is
very remote and has no nearby anthropogenic aerosol sources apart from
scientific operations near the site; when air masses blow from the direction
of the scientific camp, data are marked as contaminated and are not included
in this analysis. The inlet at Summit has a 2.5
The Tiksi Hydrometeorological Observatory in Yakutsk, Russia, was formed
through a collaboration between the Russian Federal Services for
Hydrometeorological and Environmental Monitoring (Roshydromet), NOAA, FMI,
and NSF. Though there has been a meteorological observatory at this location
since the 1930s, the new international site was established in 2009. The
site is located in northern Siberia in the Sakha Republic of Russia, just
500 m from the coast of the Laptev Sea and
The Zeppelin Mountain observatory is located on a small mountain at 475 m a.s.l.,
just south of the small research village of Ny-Ålesund (30–150
inhabitants, depending on time of year) on Svalbard in Norway. The
monitoring station is owned by the Norwegian Polar Institute and operated by
the Norwegian Institute for Air Research (NILU), and the most recent version
of the station building was constructed in the year 2000. The site is
typically located above the inversion layer and thus measures air masses
with minimal contamination. Aerosol instruments sample from an inlet line
that reaches room temperature (
Although monitoring networks offer scientists an opportunity for regional cross-station analyses of aerosol seasonality and climatologies, comparing data across monitoring sites requires caution. Care must be taken to ensure data are measured, edited, and corrected using comparable high-quality methods. Moreover, comparing the same aerosol property measured by different instrument types or models necessitates extra attention. This section describes the data and steps taken to ensure comparability of those data for this analysis.
All six sites in this analysis have scattering measurements for years 2012–2014 from an integrating nephelometer (TSI model 3563) measuring at three wavelengths (450, 550, 700 nm). Corrections to the raw scattering coefficient measurements are necessary to account for light source and angular non-idealities, and the correction methods described in Anderson and Ogren (1998) were used to correct the scattering coefficient data presented here.
In this analysis, absorption data are available from Aethalometers as well as other co-located filter-based absorption instruments (i.e., Continuous Light Absorption Photometer (CLAP), PSAP, and/or MAAP) at each observatory. The Magee Aethalometers are the only common absorption instrument among the six stations presented here, and this paper synthesizes the absorption data from Aethalometers across the Arctic. The Aethalometer data are corrected using the new Arctic-specific Aethalometer correction scheme presented by Backman et al. (2017). We use the reference co-located absorption instruments to gauge whether the corrected Aethalometer data are similar to what is expected for absorption coefficient values from other absorption measurements at the stations. The different co-located absorption instruments and Aethalometer data are described below.
Co-located reference absorption data at ALT are from a Radiance Research three-wavelength (467, 530, 660 nm) particle soot absorption photometer (PSAP-3W) and at ZEP are from a custom-built one-wavelength (525 nm) particle soot absorption photometer (PSAP-1W). The PSAP collects aerosol particles on a filter and relates the change in light transmission through the filter over time to the absorption coefficient of the deposited aerosol. PSAP data are corrected using the correction schemes from Bond et al. (1999) and Ogren (2010) to adjust for multiple-scattering effects, filter loading, apparent absorption, flow bias, spot size bias, and spectral scattering. Correcting for apparent absorption requires concurrent measurements of aerosol light scattering, which are available from TSI nephelometers at all six stations.
Co-located absorption data at BRW and SUM were measured using a CLAP at three wavelengths (467, 528, 652 nm). The CLAP is an instrument designed and built by NOAA that is based on the PSAP design, except that it samples consecutively on eight filter spots on one large 47 mm filter, as opposed to the one spot available on the 10 mm PSAP filter. The CLAP's multi-spot functionality enables it to run unattended for 8 times longer than the PSAP, making it ideal for remote, less frequently visited locations (Ogren et al., 2017). The CLAP data are corrected the same way as the PSAP using Bond et al. (1999) and Ogren (2010) corrections.
Comparison of monthly medians (computed from hourly data) and
corrected Aethalometer absorption (light blue) and absorption measured by
co-located absorption instrument (orange).
PAL and TIK co-located reference absorption data are from a Thermo Fisher Scientific MAAP at one wavelength (637 nm) (Müller et al., 2011). The MAAP is a filter-based absorption instrument that measures filter transmittance as well as backscattered light at two angles (Petzold and Schönlinner, 2004). The backscattering measurements at different angles allow the instrument to account for multiple-scattering and apparent absorption effects. Due to the low concentrations in the Arctic, no post-processing corrections are needed (Hyvärinen et al., 2013).
In addition to the co-located absorption measurement, all monitoring
stations have absorption data collected from some model of the Magee
Aethalometer. During 2012–2014, five of the stations – ALT, BRW, ZEP, PAL,
and TIK – operated a seven-wavelength (370, 470, 520, 590, 660, 880, and 950 nm)
Aethalometer AE31, while SUM operated a one-wavelength (880 nm) Aethalometer
AE16. The Aethalometer measures light transmitted through a filter on which
particles are deposited and interprets the change in transmittance, or the
attenuation of light through the filter, as the aerosol light absorption,
which the instrument reports as an atmospheric concentration of equivalent
black carbon (eBC) (Petzold et al., 2013) particles using a mass absorption
cross section of black carbon. There are known artifacts associated with
measuring absorption coefficients on the Aethalometer filter tape, including
multiple scattering by filter fibers, scattering by aerosol deposited on the
filter, and decrease in sensitivity with increased filter loading. Many
Aethalometer correction schemes exist that try to account for one or all of
these artifacts (e.g., Collaud Coen et al., 2010; Drinovec et al., 2015),
including GAW recommendations for the AE31 contained in GAW report 227
(
Backman et al. (2017) present an Arctic-specific multiple-scattering
enhancement factor,
The Aethalometer absorption data corrected with the Backman et al. (2017)
correction factor are compared to absorption coefficients from the
co-located absorption instruments to ensure that the corrected Aethalometer
data are similar to absorption coefficients that are measured by other
absorption instruments at the site. Figure 2 shows a time series of monthly
median corrected Aethalometer data and co-located absorption data from
2012 to 2014 at each site. Data are adjusted to a co-located absorption
instrument wavelength, except for SUM data, where the co-located absorption
data are adjusted to the wavelength of the one-wavelength Aethalometer (880 nm).
Wavelengths of the data in Fig. 2 are 467 nm at ALT, 467 nm at BRW,
637 nm at PAL, 880 nm at SUM, 637 nm at TIK, and 525 nm at ZEP. Note that
the
Although agreement between Aethalometer-measured absorption and co-located instrument absorption is imperfect and variable among stations, corrected Aethalometer data from all sites are utilized in the remainder of this paper for analyses of absorption coefficients at all six Arctic monitoring stations. Using Aethalometer measurements at each location, rather than three different types of co-located reference instruments (PSAP, CLAP, and MAAP), eliminates issues with comparing data from different measurement techniques across stations. Furthermore, despite the differences in instrument agreement highlighted above, much of the difference in Aethalometer and co-located reference absorption values falls within combined instrumental uncertainties, as discussed later in this section.
Measurements from all instruments used in the analysis are reported at
standard temperature and pressure (STP;
Quality assurance and quality control procedures were applied to the datasets at all six stations. Station scientists looked at each week of data individually to determine validity of the measurements. Additionally, there was a second stage of data review by the authors of this paper to double-check the data quality. During time periods where instruments appeared to be malfunctioning, or data were obviously influenced by local pollution (i.e., not representative of regional aerosol), data were invalidated or marked as contaminated. This helps ensure that data included here are representative of regional Arctic aerosol. At the sites in the study, measurements of absorption and scattering are made sub-hourly (data frequency: 1–5 min), though all data used in the analysis are hourly averages to improve the signal-to-noise ratio at the clean Arctic locations. All data used in this analysis are archived and accessible from the EBAS database operated by the NILU.
The variables analyzed here include extensive aerosol optical properties
that depend on aerosol amount – absorption (
SAE describes the wavelength dependence of the aerosol light-scattering
coefficient and is inversely related to aerosol size such that large
aerosols have small SAE values and vice versa (Delene and Ogren, 2002). SAE
is calculated using Eq. (2):
Seasonality of the aerosol light-scattering coefficient (
SSA is the ratio of scattering to extinction, as given in Eq. (3), and is
indicative of aerosol darkness such that white aerosols (e.g., sea salt)
have high SSA values and dark aerosols (e.g., black carbon) have low SSA
values. SSA is calculated using Eq. (3):
Uncertainties in PSAP- and CLAP-measured absorption coefficient measurements
come from instrumental noise, unit-to-unit variability, and instrument
calibration, with a total measurement uncertainty of
The seasonality of aerosol light scattering (
Statistics of aerosol optical properties at six Arctic monitoring
sites, including geometric means, medians, and interquartile spread of
absorption coefficient (
The scattering coefficient boxplots for each station in Fig. 3 show that the spread of scattering data is generally greatest during months when the scattering coefficient values are highest at each station. In other words, at ALT, BRW, TIK, and ZEP, the winter months have the largest range of scattering values (and the largest median scattering values), while the summer months have a smaller range of scattering values (and also the lowest median scattering values). This indicates larger day-to-day aerosol variability during the Arctic haze season at these sites. PAL and SUM see a larger spread of the scattering data during summer when scattering values are the highest. Episodic long-range transport of biomass burning aerosol (i.e., smoke), long-range transport of anthropogenic aerosol from Europe, and regional biogenic emissions are likely contributing factors to the higher summer scattering values and spread of the data at these stations (Stohl et al., 2006a, 2007; Hyvärinen et al., 2011). Other contributing factors likely include long-range transport of anthropogenic aerosol from Europe as well as biogenic emissions (Hyvärinen et al., 2011; Tunved et al., 2006). In addition, at PAL, there is increased contribution from continental air masses during the summer, which contribute to the higher scattering values (Aalto et al., 2002; Asmi et al., 2011).
Seasonality of aerosol light absorption coefficient (
Figure 4 shows monthly median values of the aerosol light absorption coefficient
(
Seasonality of single-scattering albedo (SSA) at all sites. Large
plot shows monthly medians of hourly average SSA at 550 nm at each station;
subplots below show boxplots of SSA at individual sites, with horizontal line
at the median, edges of the box at 25th and 75th percentiles, and
whiskers at 5th and 95th percentiles. Note that
SSA values show seasonality at all of the Arctic sites. Figure 5 displays monthly median values of SSA, as well as boxplots of SSA for all months and all sites. ALT has relatively constant SSA values throughout most of the year, though SSA drops during July, coincident with large variability in SSA values as seen in the ALT boxplot. The SSA values at BRW are highest in the fall (September and October) and are otherwise fairly consistent the rest of the year, with the largest spread in SSA during months other than September and October. SSA values at BRW could be highest in September and October due to low sea ice extent, more open ocean, and thus the potential for more sea salt aerosol in the area (May et al., 2016). Figure 10 lends evidence for this, and it is discussed later in the paper. The multi-year annual average of SSA at BRW was found to be 0.960 (see Table 2), which agrees with the SSA averages of 0.96 presented for BRW data from 1988 to 1993 in Bodhaine (1995) and 1997 to 2000 in Delene and Ogren (2002). PAL has higher SSA values in the summer and lower SSA values in the winter. Aalto et al. (2002) find that there is an increased contribution from continental air masses in the summer at PAL. Lihavainen et al. (2015) show that SSA in summer increases especially in continental air masses, although it is the highest throughout the year in marine air masses. The high SSA in summer is related to increasing biogenic contribution and decreasing contribution from anthropogenic sources, such as residential wood burning. SUM has similar SSA values throughout the year, except for when SSA drops to a median of 0.890 in September – quite a bit lower than the annual median SSA of 0.954. Much of the increased summer operations are winding down at SUM around September, and the related increase in flights and transportation activities at this time could contribute to the lower SSA value during September. However, no instances in the data suggest contamination spikes that need removal; rather, we speculate that the increased anthropogenic activity at SUM at this time might contribute to a darker background aerosol. TIK has the most pronounced seasonal cycle in SSA, with median values of SSA around 0.860 in the winter and higher SSA median values around 0.960 during the summer. TIK measures the darkest aerosol of all six Arctic stations during the winter. We speculate this could be due to an inversion layer trapping regional combustion aerosol produced from anthropogenic activities, energy production, and transport, mainly in the town of Tiksi and nearby villages. ZEP does not have a very distinguishable seasonality in SSA, though SSA values tend to be slightly lower during the Arctic haze season. The boxplots of SSA at ZEP indicate large variability in the SSA data at this station.
Seasonality of scattering Ångström exponent (SAE) at all
sites. Large plot shows monthly medians of hourly average SAE at the
450 and 700 nm wavelength pair at each station; subplots below show boxplots of
SAE at individual sites, with horizontal line at the median, edges of the box
at 25th and 75th percentiles, and whiskers at 5th and
95th percentiles. Note that
Seasonality of aerosol asymmetry parameter (
SAE for the 450 and 700 nm
wavelength pair is indicative of particle size and has a seasonal signature
at only some of the Arctic stations (Fig. 12). At ALT, the variability in SAE values is
highest in the summer and fall months, suggesting that the site measures a
variety of particle sizes during this time. However, the monthly median SAE
does not show substantial change throughout the year. BRW does have
seasonality in SAE, with lowest SAE values (larger particles) during the late
summer and early fall, and higher SAE values in the spring (smaller
particles). This same SAE seasonality at BRW was also observed in previous
studies (Bodhaine, 1983; Delene and Ogren, 2002), and one study offers an
explanation as to this seasonality with observations of an increase in sea salt
when the sea ice melts in summer months (Quinn et al., 2002). PAL has a
different seasonality, with highest SAE values in the summer and lowest SAE
values in the winter and early spring, which agrees with findings from
Aaltonen et al. (2006) and Lihavainen et al. (2015). The statistics of SAE in
Table 2 show an average SAE of 1.66 at PAL, which is close to the average of
Systematic variability of median aerosol optical properties:
The variability of the asymmetry parameter,
The systematic variability of aerosol optical properties refers to how aerosol parameters covary with each other. Analysis of the systematic relationships between aerosol optical properties is useful because it can provide insight to aerosol sources and atmospheric processes (Andrews et al., 2011; Toledano et al., 2007) and can also be a good metric for comparing consistency between aerosol models and measurements.
The systematic variability plots shown here were created by binning the
hourly averages of aerosol light-scattering coefficient values into 2 Mm
The absorption coefficient varies with the scattering coefficient almost linearly,
such that absorption increases as scattering increases as shown in Fig. 8a.
One interpretation of this linear relationship between these
scattering and absorption coefficients is that the scattering and absorbing
aerosols are coming from the same sources and are subject to similar
removal processes during transport to the site. This is consistent with
systematic variability analysis from Andrews et al. (2011) that looked at
data from mountain sites. Delene and Ogren (2002) also show this systematic
variability for BRW, over the same scattering range (0–20 Mm
Single-scattering albedo varies with scattering such that the lowest
scattering coefficient bins are accompanied by relatively low SSA values,
and SSA values plateau with higher scattering values – see Fig. 8b. This
finding follows the same pattern but with a much weaker dependence than what
was found for mountain sites in Andrews et al. (2011) and shows a much
weaker relationship than what was found for continental North American sites
in Sherman et al. (2015). It should be noted that comparisons with
systematic variability relationships for other site types are difficult
since this Arctic analysis only looks at scattering from 0 to 20 Mm
The scattering Ångström exponent varies with scattering in diverse ways at the six Arctic stations, as indicated in Fig. 8c. At sites like ALT, SUM, and TIK, SAE does not vary much with changes in scattering. BRW generally shows decreases in SAE (or increases in particle size) as scattering increases. Delene and Ogren (2002) show that the aerosol particles at BRW tend to be largest (lowest SAE) and whitest (highest SSA) during the summer (lowest scattering values), which they attribute to the contribution of marine aerosol when the sea ice melts. Chemical analysis has supported this conclusion (Quinn et al., 2002), though the systematic variability plots shown here do not provide the means to analyze this seasonality. ZEP shows distinctly different systematic variability from BRW, in that SAE increases (decreasing particle size) as scattering increases. The asymmetry parameter provides another means of investigating changes in particle size distribution with loading, and the plot of scattering vs. asymmetry parameter and the associated discussion are included in the Supplement.
Figure 8d shows that SAE also varies with SSA. At ALT, BRW, and ZEP, SAE
decreases as SSA increases. This indicates that the more scattering
particles are typically larger at these sites (e.g., sea salt), and more
absorbing particles are typically smaller (e.g., black carbon). There are
not enough data that meet the standard error threshold to detect systematic
variability in these properties at TIK. PAL and SUM do not show substantial
systematic variability in these optical parameters, likely due to the 2.5
Back-trajectory analyses are widely used to investigate the effect of air mass pathway on atmospheric constituents measured at a particular place (Fleming et al., 2012). The trajectory method involves calculating air parcel movement from the monitoring site back in time to yield the back trajectory of the parcel (Draxler and Hess, 1998). Here, individual 7-day back trajectories computed for each of the six Arctic sites are overlaid and colored by frequency of back-trajectory occurrence in each grid box to create a density plot of air mass history for each station.
In this work, the air mass back-trajectory analysis was conducted using the
NOAA Hybrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT)
version 4.9 (Draxler and Hess, 1998; Stein et al., 2015). The HYSPLIT model
was run for 7-day back trajectories, using an ensemble method. The ensemble
method offsets the meteorological grid by one grid point in the horizontal
and 1 % of the surface pressure in the vertical, which produces 27
back trajectories for possible offsets in the horizontal and vertical, thus
accounting for uncertainties in the gridded meteorological data. The
meteorological data used for the trajectories were from the NCEP/GDAS dataset with
a 1
Seven-day back trajectories at each of the six Arctic stations, separated by summer (May–October) and winter (November–April) months. Colors represent frequency (units of hours per 2 years) at which an air parcel travels over that region before arriving at the station, in other words, residence time of air in that location. These plots show data from all trajectory altitudes. Black dots show station location.
Figure 9 shows density plots of each 7-day back-trajectory path computed at each station over the period of interest (2012–2014), colored by frequency at which the air mass passed through the given grid cell. Regions colored in red represent regions through which air masses most frequently traveled en route to the monitoring station, and regions colored in blue represent areas through which an air mass passed least frequently en route to the monitoring station. All trajectory altitudes are included in plots in Fig. 9.
For all measurement sites, air masses arriving at the site obviously pass most frequently through regions closest to the stations. The differences between summer and winter back trajectories at each site are subtle and do not reflect the large seasonality observed in aerosol optical property measurements throughout the Arctic. This is consistent with similar back-trajectory frequency analyses at ALT (Sharma et al., 2006; Huang et al., 2010). This could be because the wide range of synoptic-scale weather patterns averaged into 3 years of back-trajectory data obscure seasonality in large-scale air mass paths. One feature that is evident from Fig. 9 is that SUM does not seem to have the same air mass origin as the other sites. Even the closest station, ALT, does not overlap much with calculated source areas for SUM. This feature is even clearer when only trajectory altitudes below 500 m a.g.l. are considered. This supports the earlier argument that, due to the altitude and location of SUM on top of the Greenland ice shelf, the aerosol arriving at the stations is very different compared to the other sites that are almost exclusively coastal. The strong seasonality observed in the aerosol optical properties at each of the Arctic sites is likely not due to large changes in air mass back trajectories from season to season. If the seasonality of the aerosol parameters is not described by differences in air mass origin, then we speculate that the aerosol sources (both natural and anthropogenic) differ in type and magnitude from season to season and may explain the temporal variability of aerosols in the Arctic. This notion is supported by previous studies (Eleftheriadis et al., 2009; Asmi et al., 2016; Wang et al., 2014; Sharma et al., 2013). Alternatively, it is possible that much longer back trajectories would elucidate additional information on seasonal differences in air mass origin for long-distance aerosol transport to the Arctic (Qi et al., 2017). For example, work by Hirdman et al. (2010) uses 20-day back trajectories from FLEXPART and suggests stronger seasonal differences in aerosol transport pathways than were found here. Using much longer back-trajectory calculations in this study would, however, also be associated with much greater uncertainties in the spatial domain, which is why the trajectory calculations were restricted to 7 days.
Percent of air mass residence time during the past 7 days spent above different land types before arriving at monitoring station for each month of the year. Green represents land (with no distinction between snow-covered and bare land areas), light blue represents sea ice, and dark blue represents open water.
For further exploration of why aerosol sources (rather than transport) might differ in type and magnitude from season to season, Fig. 10 affords insight into how the land type over which an air mass travels might affect the aerosols within it. Figure 10 shows the percent of air mass residence time spent above different land types before arriving at each monitoring station for each month of the year. The data used for sea ice extent came from the National Snow and Ice Data Center's Sea Ice Index dataset (Fetterer et al., 2015). The green bars represent land (with no distinction between snow-covered and bare land areas), light blue bars represent sea ice, and dark blue bars represent open water. There is a clear seasonality in land type over which air masses travel before arriving at each measurement site. At all sites except SUM, air masses travel more over open water during the summer when sea ice has melted. This provides a source for sea salt and other marine aerosol during the summer that is much less likely at other times in the year. The result that the same source region overlaps with open ocean in summer and sea ice in winter, and thus yields different aerosol, is supported by similar findings from Shaw et al. (2010). TIK, PAL, and SUM are similar in that most of the air mass residence time is spent above land at all times of the year, but especially so in winter. ALT, ZEP, and BRW are similar in that the air masses arriving at these stations spent more time, compared to the other sites, over sea ice and much less time over land. This could explain why ALT, ZEP, and BRW have very similar seasonality of aerosol light-scattering and absorption coefficients, while TIK, PAL, and SUM have different seasonality that may be indicative of varying land-based aerosol sources. More work is needed, using chemical analyses or footprint analyses, to better understand how air mass transport contributes to the different aerosol seasonality at each of the six Arctic sites.
Seasonal cycles of aerosol optical properties from six Arctic monitoring stations have been presented here. Aerosol optical properties were derived from common absorption and scattering instruments (Aethalometers and nephelometers, respectively) at the stations, were evaluated and corrected under common quality control procedures, and were presented at standard temperature and pressure and low relative humidity to ensure high quality and comparability of data across stations.
The extensive aerosol optical properties, dependent on amount of aerosol,
showed strong seasonality at all of the Arctic sites analyzed here. The
magnitude and variability of the aerosol light-scattering coefficient vary
substantially among stations, with SUM measuring the lowest annual mean
scattering coefficients (1.74 Mm
The intensive aerosol optical properties, which are independent of aerosol amount, also show strong seasonality at all six Arctic stations. Furthermore, quite high SSA values at all stations are evident in our data. The range of annual mean single-scattering albedo values at the sites is from 0.909 at PAL to 0.960 at BRW. The annual mean scattering Ångström exponent values range from 1.04 at BRW to 1.80 at SUM. The annual mean aerosol asymmetry parameter values range from 0.57 at ALT to 0.75 at SUM. The seasonalities of these variables suggest that aerosol source and removal mechanisms are likely different from month to month at a given site and from site to site throughout the Arctic.
Systematic variabilities of the aerosol optical parameters measured in the Arctic provide insight into atmospheric processes near the monitoring stations. Generally, absorption coefficients increase as scattering coefficients increase at all of the sites. However, the ratio of absorption to scattering is different across sites and aerosol loadings, with TIK and ALT showing higher absorption-to-scattering ratios at high aerosol loadings, and PAL showing higher absorption-to-scattering ratios at low aerosol loadings compared to the other stations. Single-scattering albedo is low at low loadings for all of the six Arctic sites, and SSA increases with increasing scattering for most sites. TIK is an exception to this observation, since darker aerosol (low SSA) is measured at higher scattering coefficients, which suggests absorbing aerosol (e.g., black carbon) may be associated with high-aerosol-loading events (e.g., anthropogenic emissions, Siberian wildfires). Our findings of generally higher aerosol absorption and lower SSA for both TIK and PAL during winter could suggest a closer proximity to anthropogenic activities, which is supported by their geographic locations since they are both continental Eurasian locations – closer to forest fires, long-range transport, and regional emissions.
Back-trajectory analysis showed little evidence of seasonality in air mass origin between winter and summer months. The analysis further strengthens the observation that SUM is different from the other stations because other stations seem to receive little air from the same areas that SUM does. Data on sea ice combined with air mass movement indicated that TIK and PAL receive the most continental air masses, whereas BRW, PAL, and ZEP are the stations with the potential to be most influenced by marine aerosol.
A persistent and important theme in the findings of this paper is that aerosol optical properties vary widely with season at any individual site, and they vary widely from station to station throughout the Arctic. This result is important, since it means that the Arctic cannot be treated as a uniform region, spatially or temporally, in climate models or in remote-sensing retrieval algorithms. Rather, the wide spatiotemporal variability of aerosol in the Arctic needs to be considered in order to properly represent the climate of this sensitive region.
Data used in this article are archived and accessible from the EBAS database
operated at the Norwegian Institute for Air Research (NILU) (
The supplement related to this article is available online at:
LS created the dataset, completed data analysis, created Figs. 1–8, and wrote most of the manuscript. JB corrected the Aethalometer data, contributed the trajectory analysis and Figs. 9 and 10, and wrote most of the trajectory analysis section. EA helped write the introduction section and helped guide the data analysis. JO and EA helped guide the data analysis and interpret results. SS and TU provided organizational support and helped guide data analysis. TU, SS, KE, SV, MB, PT, and AJ all provided station data and provided input for interpretation of results at the individual stations.
The authors declare that they have no conflict of interest.
Thank you to all of the station technicians at these Arctic monitoring sites who work in difficult Arctic conditions to help acquire the data presented here. The authors would like to acknowledge the International Arctic System for Observing the Atmosphere (IASOA) aerosol working group for coordination of the project and contribution of expertise to this analysis. Data management is provided by the WMO Global Atmosphere Watch World Data Centre for Aerosol. This project has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement no. 654109 (ACTRIS). The Finnish Meteorological Institute acknowledges the Academy of Finland project “Greenhouse gas, aerosol and albedo variations in the changing Arctic” (project number 269095); the Novel Assessment of Black Carbon in the Eurasian Arctic: From Historical Concentrations and Sources to Future Climate Impacts (NABCEA) (project number 296302); the Academy of Finland Centre of Excellence program (project number 307331); and EU H2020 project INTAROS (project ID: 727890) for financial support. Funding from the NOAA Climate Program Office provided partial support for data analysis and measurements at Barrow and Summit. The authors would like to thank the staff of the Canadian Forces Service for maintenance of the Alert station. The light-scattering measurements at Alert were initiated by Richard Leaitch. Edited by: Nikos Hatzianastassiou Reviewed by: three anonymous referees