You are on page 1of 22

Atmospheric Environment 244 (2021) 117882

Contents lists available at ScienceDirect

Atmospheric Environment
journal homepage: http://www.elsevier.com/locate/atmosenv

Simulation of long-term direct aerosol radiative forcing over the arctic


within the framework of the iAREA project
K.M. Markowicz a, *, J. Lisok a, P. Xian b
a
Institute of Geophysics, Faculty of Physics, University of Warsaw, Warsaw, 02-093, Poland
b
Naval Research Laboratory, Marine Meteorology Division, Monterey, USA

H I G H L I G H T S

• Anthropogenic, biogenic, and sea salt particles are the most important aerosols in the Arctic.
• The total aerosol radiative forcing in the Arctic (>70.5oN) is − 0.4 W/m2 for all-sky conditions.
• Anthropogenic aerosol radiative forcing is almost zero for all-sky conditions.
• Clouds reduce clear-sky aerosol radiative forcing by about 40%.

A R T I C L E I N F O A B S T R A C T

Keywords: This paper presents the climatology of aerosol optical properties and radiative forcing over the Arctic obtained
Aerosol within the framework of the iAREA (impact of absorbing aerosols on radiative forcing in the European Arctic)
Absorbing aerosol project. The presented data were obtained from the Navy Aerosol Analysis and Prediction System (NAAPS) and
Radiative forcing
the Fu-Liou radiative transfer model. NAAPS was used to simulate particle concentration and aerosol optical
Aerosol optical depth
depth (AOD) at 1 ◦ × 1 ◦ spatial resolution. Direct aerosol radiative forcing (ARF) was calculated for clear-sky
Single scattering albedo
and all-sky conditions based on NAAPS reanalysis (with AOD assimilation) and satellite observations of sur­
face and cloud properties. Long-term data (2003–2015) from NAAPS show that anthropogenic and biogenic
aerosol, as well as sea salt, make the most important contribution to total AOD (35 and 30%, respectively).
However, smoke (15%) and mineral dust (20%) cannot be neglected, especially during spring and summer.
Results of numerical simulations indicate mean shortwave (SW) ARF for the whole Arctic (>70.5oN) at the
Earth’s surface to be − 4 W/m2 for clear-sky and − 1.3 W/m2 for all-sky conditions, and at top of the atmosphere
(TOA) − 1.3 W/m2 and -0.4 W/m2, respectively. TOA ARF for anthropogenic and biogenic particles is only − 0.1
W/m2 for clear-sky and almost zero for all-sky conditions. For smoke and dust particles, SW ARF is very similar
for both Earth’s surface and TOA, as well as for clear-sky and all-sky conditions. For sea salt, SW ARF is the same
at the surface and at TOA: 0.6 W/m2 for clear-sky and − 0.3 W/m2 for all-sky conditions, because of negligible
solar absorption. Cloud cover reduces surface cooling (direct clear-sky SW ARF) by a factor of 40% and shifts
TOA SW ARF towards positive values.

1. Introduction of sea ice (Willis et al., 2018). The warming is suggested to be forced
mainly by greenhouse gases, and additionally by absorbing aerosols
The Arctic region is currently experiencing unprecedented climate (Bond et al., 2013). However, understanding the variability of the Arctic
change. The warming rate in this region is 2–3 times the global average climate is still a challenge due to strong interactions between local and
(IPCC, 2018). A clear manifestation of this is the dramatic decline in the non-local drivers and feedbacks (Struthers et al., 2011; Abbatt et al.,
extent of Arctic sea ice observed in recent decades (Barry, 2017; Over­ 2019). Previous studies used model simulations to investigate how
land and Wang, 2013; Comiso et al., 2008). It is uncertain what factors aerosols affect the Arctic climate through the effects of scattering and
drive the changes observed in Arctic surface temperatures and the extent absorption on radiation, clouds, and surface ice/snow albedo (Clarke

* Corresponding author.
E-mail address: kmark@igf.fuw.edu.pl (K.M. Markowicz).

https://doi.org/10.1016/j.atmosenv.2020.117882
Received 30 October 2018; Received in revised form 27 July 2020; Accepted 23 August 2020
Available online 28 August 2020
1352-2310/© 2020 Elsevier Ltd. All rights reserved.
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

and Noone, 1985; Flanner et al., 2007; Dumont et al., 2014). Ren et al. between 1980 and 2010, TOA-negative ARF in the Arctic decreased from
(2020) report the aerosol produces an Arctic surface warming of +0.297 − 0.67 ± 0.06 W/m2 to − 0.19 ± 0.05 W/m2, yielding a net TOA ARF
K during 1980–2018, explaining about 20% of the observed Arctic increase of 0.48 ± 0.06 W/m2and a net warming of 0.27 ± 0.04 K at the
warming during last four decades. Quinn et al. (2008) found that Arctic surface. However, Yang et al, (2014), using a climate model with
anthropogenic aerosol warms the Arctic atmosphere and cools the sur­ different radiative transfer model and aerosol optical properties of
face in all seasons. Shindell and Faluvegi (2009) reported that changes in aerosols, found no significant response in the average Arctic surface
aerosols from 1976 to 2007 contributed 1.09 ± 0.81 ◦ C to the observed temperatures to the 1975–2005 trends in sulphates and BC.
Arctic surface temperature increase of 1.48 ± 0.28 ◦ C. However, other In this study, we provide an analysis of the climatology of aerosol
numerical simulations indicated anthropogenic aerosols yielded a optical properties and ARF in the Arctic within the framework of the
negative forcing over the Arctic. Breider et al. (2017), based on the iAREA (impact of absorbing aerosols on radiative forcing in the Euro­
GEOS-Chem model and the Fu-Liou radiative transfer code, reported pean Arctic) project based on the Navy Aerosol Analysis and Prediction
shortwave (SW) aerosol radiative forcing (ARF) of − 0.19 ± 0.05 W/m2 System (NAAPS) reanalysis, which includes aerosol optical depth (AOD)
at the top of the atmosphere (TOA), which reflects the balance between data assimilation. This is just to show the difference between this study
sulphate cooling (− 0.60 W/m2) and black carbon (BC) warming (+0.44 and other modeling studies based on models which did not apply AOD
W/m2). Surface ARF from anthropogenic particles averaged over the assimilation. Section 2 includes the description of the model and the
Arctic during 2005–2010 was − 1.20 W/m2, which was roughly equiv­ databases, followed by the validation of the model and the sensitivity
alent to the sum of the cooling from sulphate aerosols and BC warming. analysis of the radiative transfer code’s input parameters, as well as the
These results stand in contrast to those obtained by Quinn et al. (2008), long-term mean aerosol optical properties. In Section 3 we discuss their
where the direct ARF was estimated from 0.08 W/m2 during winter and long-term mean values for the whole Arctic spatial distribution, as well
autumn to 0.92 W/m2 in spring. The significant differences in the as the annual variability for selected Arctic regions. Conclusions are
simulated impact of the anthropogenic aerosols were due to BC, which summarised in section 4.
in Quinn et al. (2008) is the dominant constituent of the total ARF. The
effect of BC on the climate system, especially in the Arctic region, is 2. Model description
highly uncertain (Bond et al., 2013). BC aerosols in the Arctic can per­
turb the radiation balance directly, through the absorption and scat­ ARF in the Arctic is computed based on an off-line radiative transfer
tering of radiation, or through the reduction of high surface albedo (Dou model (RTM), which was interfaced with several databases including
and De Xiao, 2016). ARF by BC can also result from aerosol-cloud in­ aerosol transport model results and satellite observations. For this pur­
teractions (Twomey, 1977) and from semi-direct effect (Koch and Genio, pose, the 200,503 version of the Fu-Liou code (Fu and Liou, 1992, 1993),
2010). Sand et al. (2013) showed that increased BC forcing in the Arctic which employs the delta 2/4 stream solver and includes six shortwave
atmosphere reduces the surface air temperature in the Arctic. This effect (SW, λ < 4 μm) and 12 longwave (LW, λ ≥ 4 μm) spectral bands, was
is due to the combination of the weakening of the northward heat used. The Fu-Liou code is robust and is therefore suitable for computing
transport, caused by a reduction in the meridional temperature gradient, ARFs on a global scale. At the same time, it has sufficiently good accu­
and the dimming at the surface. Moreover, model simulations indicated racy for such global applications, as indicated by the inter-comparison,
that atmospheric BC forcing outside the Arctic may be more important with more precise RTMs performed by Myhre et al. (2009). The Fu-Liou
for Arctic climate change than the forcing in the Arctic itself (Sand et al., RTM was interfaced with various databases of optical properties (aero­
2013). sols, clouds, and the Earth’s surface) and meteorological data to
Comparison of different models (AeroCom Phase II) showed that the compute the radiation fluxes on a regular grid with a horizontal reso­
impact of BC aerosols on the global radiation balance is not well con­ lution of 1 ◦ × 1 ◦ and 25 vertical levels, ranging from the surface to the
strained (Samset et al., 2013). Both BC burden and BC ARF were well level of about 18 km. The datasets were defined at each grid box by
spread (Myhre et al., 2013). Twelve global aerosol models showed that simple interpolation (in time and space) of data from databases
at least 20% of the presented uncertainty in the modelled BC direct ARF described below. In this study, the diurnal mean (24 h) ARF was
was due to diversity in the simulated vertical profile of BC mass (Samset computed from diurnal cycles. A higher time resolution was required for
et al., 2013). In addition, a significant fraction of the variability came the daytime than for the night-time to ensure sufficient accuracy for the
from high altitudes, especially in the Arctic, where BC is transported mean diurnal ARF estimates. For this purpose, the RTM was run with a
from middle latitudes. The models might have overestimated BC con­ time step of 20 min for the cases in which the sun was above the horizon,
centrations in the Arctic free troposphere, but the vertical profile of and of 3 h during the night.
Arctic BC is not well known due to sparse vertical measurements (Sand The aerosol optical properties used in the study come from the long-
et al., 2013). term (2003–2015) aerosol reanalysis product based on the Navy Aerosol
Much of the aerosol in the Arctic originates from various parts of the Analysis and Prediction System (NAAPS) with AOD assimilation. The
Earth (Willis et al., 2018). For example, a modeling study from Bour­ NAAPS reanalysis product includes aerosol mass concentration and
geois and Bey (2011) indicates that 59% of sulphates in Arctic tropo­ other aerosol variables associated with it, at a spatial resolution of 1 ◦ ×
sphere come from the oxidation of SO2 emitted in Siberia (19%), Europe 1 ◦ and a temporal resolution of 6 h (Lynch et al., 2016). This reanalysis
(18%), Asia (13%), and North America (9%), while anthropogenic BC utilises a modified version of NAAPS with a data assimilation module,
and biomass burning BC emitted in Siberia, Europe, Asia, and North including quality-controlled retrievals of total AOD from MODIS on the
America contribute 29, 25, 27, and 17%, respectively, to the Arctic BC Terra and Aqua satellites, as well as the Multi-angle Imaging Spec­
burden. However, due to the reduction of aerosol and precursor emis­ troRadiometer (MISR) on Terra (Zhang et al., 2006; Zhang et al., 2008;
sion in Europe, Russia, and North America in recent decades, in situ Hyer et al., 2011; Shi et al., 2014). The reanalysis version of NAAPS is
observations in the Arctic have shown a significant decline in concen­ driven by the meteorological fields of Navy Operational Global Analysis
trations of sulphate and BC aerosols since 1980 (Gong et al., 2010; and Prediction System (NOGAPS; Hogan and Rosmond, 1991). NAAPS
Hirdman et al., 2010). In the observations reported by Sharma et al. model uses NOGAPS analysis (00, 06, 12, and 18 UTC) with associated
(2013), atmospheric BC at three Arctic locations showed a reduction of 3-h forecast fields. NAAPS reanalysis uses the following variables:
40% between 1990 and 2009. However, sea salt particles are increased, topography, sea ice, surface stress, surface heat flux, surface moisture
due to the reduction of the sea ice extent in the Arctic (Struthers et al., flux, surface temperature, surface wetness, snow cover, stratiform pre­
2011). The latest data by Breider et al. (2017) showed a decrease in the cipitation, convective precipitation, lifting condensation level, cumulus
BC mass concentration of 2–3% per year over the period 1980–2010. As fractional coverage, cumulus cloud height, surface pressure, wind vec­
a result, according to their calculation for total anthropogenic particles, tor, temperature, and relative humidity. Parametrization of cloud and

2
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

planetary boundary layer processes in the NOGAPS model was previ­ factor (GF)
ously described by Teixeira and Hogan (2002). ( )− γi
1 − RH
NAAPS characterises anthropogenic and biogenic fine (ABF, GFi (RH) = (1)
1 − RHo
including sulphate, primary and secondary organic aerosols) aerosols,
dust, smoke, and sea salt aerosols. Physical processes related to sea salt
where RH and RHo are the ambient and reference (30%) relative hu­
particles for the NAAPS model were developed by Witek et al. (2007a).
midity, respectively. An empirical factor (γi) describes the humidity
The emission source function is based on wind speed at 10 m a.s.l. ac­
effect for each aerosol category. Parameter γi is defined as 0.5 for ABF
cording to Monahan’s formulation (Monahan et al., 1986). Similarly, a
particles, assuming 40% sulphate and 60% organic aerosols (Lynch
dust emission is driven by surface wind speed whenever the friction
et al., 2016), while γi is 0.63 for sulphate (Hänel, 1976), 0.18 for smoke
velocity exceeds a threshold value (6 m/s) and the surface moisture is
(Reid et al., 2005b), 0.46 for sea salt (Hegg et al., 2002; Ming and
under a critical value (Westphal et al., 1988). In addition, to estimate the
Russell, 2001), and zero for dust (Li-Jones et al., 1998). In addition, we
source function the surface erodible fraction value is used for each
assumed that the absorption coefficient is not affected by humidity.
model grid box. Emission of the biomass burning smoke in NAAPS
For the surface optical properties, the cloud-screened and quality-
reanalysis uses the FLAMBE database (Fire Locating and Modeling of
assured variables were derived from daily MODIS products of the sixth
Burning Emissions; Reid et al., 2009). To avoid changes in the geosta­
collection – M × D09CMG (Terra and Aqua). We defined the monthly
tionary constellation, which can be a challenge for time-consistency of
mean surface albedo at six bands: 0.459–0.479, 0.545–0.565, 0.62–0.67,
the smoke source function, the polar-orbiting version of FLABME was
0.841–0.876, 1.23–1.25, 1.628–1.652, and 2.105–2.155 μm. Their
used. In this case, MODIS-only fire products and emissions are applied
values were interpolated to the Fu-Liou spectral bands. The solar zenith
(Lynch et al., 2016). ABF includes sulphate, primary and secondary
angle variability was described by Markowicz and Witek (2011).
organic aerosols. Emissions for sulphate and its gaseous precursor SO2
Optical properties of clouds were used to compute all-sky radiative
are based on the Monitoring Atmospheric Composition & Climate/Ci­
fluxes. We assumed that the optical properties of clouds are horizontally
tyZen EU projects (MACCity) inventory 2005–2010 average (Granier
uniform. Monthly data of cloud properties were obtained from the sixth
et al., 2011; Diehl et al., 2012). SO2 converted from Dimethylsulphide
collection MODIS global product M × D08_M3, which uses the data from
(DMS), is derived from the monthly mean DMS seawater concentrations
Terra and Aqua satellites (Platnick et al., 2003). The dataset consisted of
from Lana et al. (2011), with the parameterization of emission fluxes at
cloud fraction, cloud optical thickness, temperature, height, and pres­
the air-sea interface from Saltzman et al. (1993). Primary organic
sure of the cloud top. The algorithm to obtain the monthly setup of
aerosol (POA) emission in NAAPS includes the major volatile organic
cloud-related variables for the Fu-Liou code was based on Markowicz
carbon (VOC) species that act as precursors for SOA (Secondary Organic
and Witek (2011), which allowed us to estimate the cloud fraction with
Aerosol). For this purpose the 2005–2010 monthly mean MACCity
randomly-overlapping altitude layers of clouds from satellite-derived
database for anthropogenic (industrial and transport) emissions of POA
cloud fraction of non-overlapping altitude layers.
and SOA precursors (Granier et al., 2011; Bond et al., 2004), as well as
The RTM requires information about vertical profiles of temperature,
biofuel data, are used. A detailed description of aerosol emissions, as
pressure, humidity, and ozone. Solar fluxes are strongly sensitive to
well as removal processes in NAAPS reanalysis, was presented by Lynch
water vapour content, so we used NOGAPS data (temperature, pressure,
et al. (2016).
and specific humidity) every 6 h to define these profiles. Because ozone
Aerosol optical properties in NAAPS are obtained from profiles of
concentration is not very responsive to SW and LW radiation fluxes, we
aerosol concentrations of four externally mixed aerosol species with no
used ozone vertical profiles from three-dimensional climatology –
interaction allowed. Profiles of the aerosol extinction coefficient, single
UGAMP (Li et al.,1995). The profiles were scaled to the total ozone
scattering albedo, and asymmetry parameter were obtained using bulk
content from MODIS M × D09CMG product.
values of optical properties that were derived from theory and obser­
vations. The predefined values of mass extinction, scattering efficiency,
and asymmetry parameter for four aerosol types were taken from the 2.1. NAAPS validation in the Arctic region
Optical Properties of Aerosol and Clouds (OPAC) database (Hess et al.,
1998). Such bulk parameters were defined for both SW and LW. The NAAPS AOD reanalysis was shown to validate well with the
Absorbing and extinction coefficients for ABF are weighted by more AErosol RObotic NETwork (AERONET) observations in general globally
absorbing aerosol particles generated by developing countries nowadays and regionally (Lynch et al., 2016). However, specific validation over
dominating global aerosol emission (Dubovik et al., 2002). Also, optical the Arctic region was not available. So we first present the validation
properties of smoke are treated similarly to ABF, according to Reid et al. results of NAAPS reanalysis for the Arctic region here. A few AERONET
(2005a, 2005b). A summary of the optical properties at 550 nm for sites with relatively long data records were selected for this purpose.
different species is available in Table 1 of Lynch et al. (2016). The effect They are Hornsund and Ny-Ålesund in Svalbard, Andenes in Northern
of humidity on the aerosol scattering coefficient at each model level is Scandinavia, Ittoqqortoormiit in Eastern Greenland, Resolute Bay in the
defined based on the Hänel (1976) equation of the hygroscopic growth Canadian Arctic, and Barrow in Alaska (Fig. 1). Table 1 shows the square
of correlation coefficient (R2), the root mean square error (RMSE), and

Table 1
Determination coefficient (R2), root mean square difference (RMSE), number of observation (N), bias between 550 nm AOD from NAAPS and from the ground-based
observations made by SP1a (Ny-Ålesund) and CIMEL (rest of the stations) sun photometers. In the case of the AERONET data, the lev. 2.0, ver. 2 product was used.
Measured AOD at 500 nm was convert to 550 nm based on 440/870 AE and next averaged to the NAAPS output (00, 06, 12, and 18 UTC) and to monthly mean values.
Last column shows long-term mean AOD (500 nm) obtained from AERONET observations.
Stations 6 h data Long-term monthly mean AERONET 550 nm AOD
2 2
N R RMSE Bias N R RMSE Bias

Hornsund (2005–2015) 1340 0.63 0.037 0.002 66 0.90 0.014 0.012 0.081
Ny-Ålesund (2003–2011) 896 0.49 0.033 0.010 49 0.93 0.021 0.019 0.067
Andenes (2003–2014) 1078 0.42 0.039 0.005 43 0.48 0.027 0.024 0.084
Ittoqqortoormiit (2010–2015) 1336 0.40 0.042 0.024 39 0.64 0.033 0.032 0.054
Resolute Bay (2004–2015) 1310 0.64 0.053 0.017 44 0.66 0.019 0.010 0.076
Barrow (2003–2015) 1344 0.51 0.067 0.008 66 0.73 0.021 0.017 0.088

3
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

the bias between NAAPS and AERONET AOD (at 550 nm) every 6 h (if Validation of NAAPS AOD components (anthropogenic and biogenic
observations were available) and for monthly means. Monthly means fine, biomass burning smoke, mineral dust, sea salt) is default due to fact
AERONET AOD were calculated from daily averaged values while for that decomposition of measured total AOD is not straightforward.
NAAPS data we used whole data set including days when AERONET Therefore it can be done during events (e.g. biomass burning or mineral
observation were not available. In the case of the 6-h data, the corre­ dust) or based on the ship measurement (sea salt). On another hand
lation coefficient (R2) varies between 0.4 and 0.63 among the sites, and model can be validated with chemical data obtained from ground-based
RMSE varies between 0.033 and 0.067. Overall, the absolute value of the measurements. Witek et al. (2007b) reports good agreement of NAAPS
mean bias is small, greater than 0.01 only for the Greenland and Reso­ surface sea salt concentration with shipboard measurements. For
lute Bay stations. Relative bias ([NAAPS-AERONET]/AERONET) example the correlation coefficient varied from 0.55 to 0.84 for different
changes between 3% and 22% for 5 stations. Only for Greenland site is experiments. On average, model overestimating the observation by
about 44%. For the monthly mean data R2 is significantly higher about 58%. It can be explained by high uncertainties of emission source
(0.48–0.93) and the RMSE smaller than the 6-h data, with the exception function. According to Jaeglé et al. (2011) and Lewis and Schwartz
of the Andenes site. For other stations, RMSE changes between 0.01 and (2004), sea salt is one of the most poorly constrained aerosols. Witek
0.03. The values of the mean bias are small and positive. For the et al. (2016) run NAAPS model with implemented different sea salt
Greenland and Andenes stations, the AOD from NAAPS is overestimated emission source functions. The annual mean global sea salt AOD varies
by about 0.02–0.03 which corresponds to relative bias of 58% and 29%, in a large range from 0.028 to 0.112. After validation to satellite data
respectively (see Fig. 2). authors recommend the Sofiev et al. (2011) SSA emission function with
Validation of NAAPS for fine (anthropogenic and biogenic fine plus the Jaeglé et al. (2011) SST scaling factor as the best choice for global
biomass burning smoke) and coarse (mineral dust plus sea salt) AOD has and regional simulations of SSA. For this parametrization the global over
been done with AERONET data. For this purpose the data obtained from ocean average sea salt AOD is 0.085 while for Monahan’s source func­
Spectral De-Convolution Algorithm (SDA) were used at 550 nm (see tion used in this study the global mean is about 0.047 (Witek et al.,
Table S1 in the supplementary material). Data analysis includes results 2011) which corresponds to about 0.062 over the ocean.
from ten high latitude AERONET stations (Hornsund, Thule, Kanger­ Validation of NAAPS SSA (or absorption AOD) and asymmetry
lussuaq, Ittoqqortoormiit, Andenes Resolute Bay, Barrow, Bonanza parameter (or scattering phase function) is limited to completely clear-
Creek, Tiksim Yakutsk) between 2003 and 2013. Fine-mode fraction sky conditions. In addition, AERONET recommendation for using ab­
(fine to total AOD ratio) of NAAPS AOD, on average, is similar to that of sorption data (Level 2.0) is defined by a AOD (440 nm) threshold of 0.4.
AERONET (65% vs. 62%). Deviation of fine-mode fraction is less than For lower AOD the uncertainty of SSA is very high (Dubovik and King,
3% for most sites, except the two sites over Siberia (6–7% higher) and 2000). Therefore such limitations prevent validation of NAAPS optical
Bonanza Creen (Alaska; 8% higher). The NAAPS bias of coarse-mode properties at high latitudes.
AOD is much smaller (0.001) compared to the bias of fine-mode AOD
(0.009). However, correlations for coarse AOD are low, suggesting the
model have difficulty producing the timing of coarse particle events 2.2. Sensitivity study of ARF with respect to input parameter uncertainty
(Xian et al., 2019). In addition, fine-mode AOD from AERONET exhibits
great temporal variability on day-to-week time-scales. NAAPS model has In this section, the sensitivity study of both SW and LW ARFs with
difficulty in capturing this large variability in fine-mode AOD. For some respect to the uncertainty of aerosol optical parameters, cloud proper­
stations NAAPS tends to overestimate in clean conditions and underes­ ties, and surface albedo is discussed. In Table 2, we present the mean
timate in highly polluted conditions. In previous research Sessions et al. value of the input parameters, which are used as a reference for the
(2015), and Xian et al. (2019) report that it is very common for global sensitivity simulations. The calculations were run for a single location in
aerosol models. two cases: one for the reference values and the second for perturbations.
The perturbations are 0.02 (22% of the regional climatological mean)
for AOD, 0.02 (2%) for SSA (which corresponds to 47% for absorbing
AOD), 0.05 (8%) for asymmetry parameter, 0.2 (27%) for AE, 0.05 (7%)
for cloud fraction, 3.6 (87%) for cloud optical depth, 0.05 (9%) for
surface albedo, and 1 km for vertical distribution of aerosol extinction.
These are reasonable perturbations based on available observations. For
AOD the perturbation is equal to mean bias between NAAPS and AER­
ONET observation. For SSA and asymmetry parameter model validation
is not available for Arctic conditions because photometer retrieval is
very uncertain at low AOD (Dubovik et al., 2002). Number of available
data from Level 2.0 for Arctic stations does not allow statistical analysis.
Therefore in this case we assumed does not allow statistical analysis
based on mean spatial standard deviation. Also for contribution of un­
certainty of cloud optical depth to ARF we used mean standard devia­
tion. For these simulations, we assumed that the vertical profiles of
aerosol extinction coefficient have a Gaussian distribution. Due to the
large variability of surface albedo, we also performed simulations for the
surface albedo of 0.2 and 0.8. Tables 3 and 4 present differences in ARF
due to perturbation input parameters as well as the reference ARF (last
line) for the surface albedos of 0.2 and 0.8, respectively. In addition,
ARF sensitivity to non-spherical particles is simulated. For this purpose,
the optical properties of randomly oriented spheroid particles are used
from Optical Properties of Aerosols and Clouds (OPAC) ver. 4.0 (Koepke
et al., 2015), as well as optical properties of spherical dust aerosol from
OPAC ver. 3.1 (Hess et al., 1998). Total ARF uncertainty (δARF) is
computed from individual uncertainty (δARFi) defined by perturbations
Fig. 1. Location of the AERONET stations used for NAAPS AOD validation. described in Table 2

4
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 2. Comparison of annual cycle of 550 nm AOD obtained from ground-based observation (red bars) and NAAPS re-analysis (blue bars) for six Artic stations
(Hornsund, Ny-Ålesund, Andenes, Ittoqqortoormiit, Resolute Bay, and Barrow).

√∑
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
ARF. Slightly lower values (5–6%) are obtained for LW ARF. The effect
δARF = δARF 2i (2)
i of a non-spherical particle is rather low, as the difference between
spherical and spheroid particle extinction coefficients, single-scattering
For the SW ARF, the most important parameter is AOD, both for albedos, and asymmetry parameters is small (Koepke et al., 2015; Fu
clear- and all-sky conditions. An uncertainty in AOD of about 20% et al., 2009). However, Kok et al. (2017) reports that the common
produces almost the same error in SW ARF over dark (0.2) and bright simplification to treat dust as a spherical particle leads to underesti­
(0.8) surface albedo (Tables 3 and 4). The SSA and asymmetry param­ mation of extinction efficiency by 20–60% for coarse aerosols Conse­
eters contribute to about 8–10% of ARF uncertainty, while AE only quently, dust direct radiative forcing would be more positive than
contributes to 2–3%. In the case of surface albedo, the uncertainty of estimated by earlier studies and likewise this study according to his
0.05 leads to an ARF error of 7–14% for darker surface and about theory. For clear-sky conditions, total SW ARF uncertainty is about 26%
20–30% for brighter. The effect is very small (below 1%) regarding and 29% at the Earth’s surface and at TOA, respectively. In the case of
aerosol altitude uncertainty. The impact of cloud uncertainty for all-sky all-sky conditions, the uncertainty increases to 29–38%. With a much
SW ARF is about 7–8% for cloud fraction perturbation of ±0.05 and brighter surface, TOA ARF uncertainty is very high and reaches 86% and
10–11% for cloud optical depth of ±3.6. Uncertainty related to mineral 63% for clear- and all-sky conditions, respectively (Table 4). For the LW
dust shape is about 6–8% for surface and TOA clear-sky and all-sky SW

5
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Table 2 where it is close to 80%. Unlike the SW ARF, which is very sensitive to
Annual mean aerosol optical depth (AOD), single-scattering albedo (SSA), surface albedo, the LW ARF is not sensitive to surface albedo.
asymmetry parameter (g), Ångström Exponent (AE), total cloud fraction, total
cloud optical depth (COD) at 550 nm, surface albedo (ALB), altitude of
maximum aerosol extinction profile (ALT), and dust non-spherical particles 2.3. Long-term mean of optical input parameters
(SPHEROID) used to estimate the ARF uncertainties.
Parameter Mean value Perturbation Relative Clear-sky diurnal direct SW ARF is a function of aerosol optical
perturbation properties, surface albedo and incoming solar radiation at TOA. LW ARF
AOD – aerosol optical 0.092 0.02 0.22
also depends slightly on the profiles of aerosol, total water vapour
depth at 550 nm content, ozone, and temperature. Fig. 3 shows long-term (2003–2015)
SSA – single scattering 0.951 0.02 0.02 annual mean AOD at 550 nm for anthropogenic and biogenic particles,
albedo at 550 nm mineral dust, biomass-burning smoke, sea salt particles, as well as total
g – asymmetry parameter 0.644 0.05 0.08
AOD obtained from NAAPS. The anthropogenic and biogenic component
at 550 nn
AE – Ångström exponent 0.75 0.2 0.27 varies from 0.016 over Greenland to 0.067 over Northern Eurasia. The
450/1000 nm variability over the Central Arctic (area around the N pole), the Barents
Cfrac - total cloud fraction 0.682 0.05 0.07 Sea, and the Greenland Sea is rather small. For the mineral dust AOD the
COD – total cloud optical 4.14 3.6 0.87 variability in the whole domain is smaller (0.011–0.028) due to the long
depth
ALB – surface albedo at 0.58 (0.2, 0.05 0.09
distance from the source areas. Smoke AOD varies from 0.007
600 nm 0.8) (Greenland) to 0.058 (Alaska). Higher values appear in places where
ALT – altitude of aerosol 1 km 1 km frequent wildfires are observed. Significantly higher values are present
extinction peak over land (North America, Asia) than over water due to the source
SPHEROID – dust non- spherical (Koepke et al.,
location. Finally, the sea salt AOD is related to the wind speed over
spherical particles particles 2015)
ocean, the ice-free area, and the distance to the ice-free ocean. There­
fore, the peak of AOD is over northern Atlantic (0.077), while the
range, the AOD and AE errors are key parameters of ARF sensitivity. minimum over Greenland for aerosol components. Similar spatial vari­
AOD and AE contribute to the uncertainty of AOD in far-infrared, which ability is shown for total AOD, which changes from 0.04 to 0.15. In all
is the most important quantity for the LW ARF. For this ARF, the un­ cases, the AOD over Greenland is a minimum, which is consistent with
certainty is about 50%, with the exception of the TOA all-sky case, the minimum long-term mean AOD over Ittoqqortoormiit among all
AERONET sites as shown in Table 2, although NAAPS has a high bias

Table 3
Annual mean absolute uncertainty of SW and LW ARF due to errors in model input parameters under clear- and all-sky conditions. Results are obtained for 69.5◦ N
latitude, flat spectral surface albedo of 0.2, and annual mean aerosol and cloud conditions (see Table 2). The penultimate row shows reference ARF, and the last row
shows its uncertainty.
Parameters Uncertainty of SW ARF [W/m2] Uncertainty of LW ARF [W/m2]

clear-sky all-sky clear-sky all-sky

SURF TOA SURF TOA SURF TOA SURF TOA

ΔAOD = 0.02 0.93 0.62 0.49 0.26 0.08 0.06 0.04 0.02
ΔSSA = 0.02 0.36 0.27 0.26 0.17 0.03 0.01 0.01 0.00
Δg = 0.05 0.42 0.43 0.21 0.21 0.04 0.04 0.02 0.01
ΔAE = 0.2 0.11 0.08 0.05 0.03 0.16 0.10 0.07 0.04
ΔALB = 0.05 0.35 0.39 0.16 0.20 0.00 0.00 0.00 0.00
ΔALT = 1 km 0.04 0.09 0.01 0.17 0.01 0.04 0.00 0.05
ΔCfrac = 0.05 0.00 0.00 0.17 0.14 0.00 0.00 0.02 0.01
ΔCOD = 3.6 0.00 0.00 0.24 0.15 0.00 0.00 0.03 0.00
SPHEROID 0.39 0.22 0.19 0.08 0.02 0.01 0.01 0.01
TOTAL 1.21 0.93 0.71 0.51 0.19 0.13 0.09 0.07
ARF [W/m2] − 4.73 − 3.23 − 2.42 − 1.35 0.38 0.26 0.17 0.09
δARF/ARF 0.26 0.29 0.29 0.38 0.49 0.50 0.51 0.77

Table 4
Same as Table 3, except for surface albedo of 0.8.
Parameters Uncertainty of SW ARF [W/m2] Uncertainty of LW ARF [W/m2]

clear-sky all-sky clear-sky all-sky

SURF TOA SURF TOA SURF TOA SURF TOA

ΔAOD = 0.02 0.19 0.23 0.12 0.23 0.08 0.06 0.04 0.02
ΔSSA = 0.02 0.12 0.76 0.12 0.57 0.03 0.01 0.01 0.00
Δg = 0.05 0.07 0.09 0.03 0.04 0.04 0.04 0.02 0.01
ΔAE = 0.2 0.02 0.01 0.01 0.01 0.16 0.10 0.07 0.04
ΔALB = 0.05 0.26 0.31 0.15 0.21 0.00 0.00 0.00 0.00
ΔALT = 1 km 0.01 0.11 0.03 0.11 0.01 0.04 0.00 0.05
ΔCfrac = 0.05 0.00 0.00 0.03 0.00 0.00 0.00 0.02 0.01
ΔCOD = 3.6 0.00 0.00 0.01 0.10 0.00 0.00 0.03 0.01
SPHEROID 0.08 0.14 0.05 0.12 0.02 0.01 0.01 0.01
TOTAL 0.36 0.88 0.24 0.68 0.19 0.13 0.09 0.07
ARF [W/m2] − 0.98 1.02 − 0.58 1.06 0.38 0.26 0.17 0.09
δARF/ARF 0.37 0.86 0.41 0.64 0.50 0.50 0.51 0.77

6
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 3. Spatial distribution of annual long-term (2003–2015) mean of (a) anthropogenic and biogenic, (b) mineral dust, (c) smoke, (d) sea salt AOD, and (e) total
AOD at 550 nm obtained from the NAAPS reanalysis.

over the site. The minimum AOD over Greenland is likely a result of lack which did not have aerosol assimilations (AeroCom phase II). However,
of local aerosol sources and the high topography, which impedes AeroCOM model median also seems to have a minimum AOD spot there,
transport and diffusion of low-altitude aerosols. On the other hand, our but with much smaller contrast to its surroundings. One of the possible
research stands in contrast to Sand et al. (2017)’s, where no significant explanation could be that the climate models could not resolve the high
AOD minimum was found over Greenland, based on 16 climate models, topography effect because of their coarse resolution (mostly very coarse

7
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

resolution, e.g. 2.5 × 2.5◦ ). Except that, spatial distributions of NAAPS 1.5 km a.s.l., over Resolute Bay it is close to 3.5 km, while the profile at
AOD for different components shows similar patterns to the AeroCom Greenland station is almost constant up to about 4 km. In the case of sea
phase II model means. However, NAAPS validation shows slightly high salt, the maximum extinction is at the surface layer. The reduction of the
bias (around 0.01) for total AOD but for AeroCom the mean bias is quite extinction coefficient with altitude is high for all stations and during all
low (~-0.03) (see Fig. 4 Sand et al., 2017). The long-term mean NAAPS seasons as sea salt aerosol fall off the atmosphere quickly because of
AOD at 550 nm for latitudes >60.5oN is significantly higher, at 0.1, than their large sizes.
that reported by Sand et al. (2017) at 0.07. Also, the contribution of the ARF is sensitive to the relative distribution of aerosol and cloud
natural particles such as mineral dust and sea salt in total AOD is about layers; therefore, fraction of AOD above clouds is calculated. For this
50–65% higher in NAAPS than in AeroCom II. purpose the profile of aerosol extinction as well as the satellite infor­
The spatial distribution of annual mean visible surface albedo (600 mation on cloud vertical distribution was used. The fraction of AOD
nm) and columnar SSA (550 nm) is presented in Fig. 4. In the Arctic, a above cloud was calculated as an integration of extinction profile above
significant variability of the surface albedo due to sea ice, glaciers, and cloud top. Fig. 6 shows the probability distribution function for AOD
snow cover is visible (Fig. 4a). For example, the surface albedo is below fraction above the highest clouds for the whole Arctic (red bars). The
0.1 about 500 km to the south of Svalbard, while it is approximately 0.8 average AOD fraction is 0.12 and the standard deviation is 0.03. Because
about 500 km to the north. This is due to the North Atlantic currents this parameter was estimated without information about cloud cover we
bringing relatively warm water from middle latitudes, allowing the sea defined a more reliable quantity (mean AOD fraction above clouds) as
surface to be free from ice cover even during winter (Polyakov et al.,

3
2017). In the case of SSA, the annual mean values were computed from τi fi
NAAPS reanalysis (Fig. 4b). The spatial variability is related to chemical < τfrac >= i=13 (3)

composition (aerosol type) and relative humidity. The highest SSA fi
(about 0.97–0.99) over the North Atlantic, Greenland, and the Barents i=1

Sea covers the region where the aerosol is dominated by sea salt parti­
where τI is the AOD fraction above the cloud “i”, and fi is the cloud
cles. Over the Central Arctic, SSA is almost constant (0.95), indicating
fraction determined at three cloud levels (low, middle, and high). As a
moderately absorbing particles. Over northern Asia (Siberia), northern
result, mean AOD fraction above clouds is 0.36 and the standard devi­
Canada, and Alaska, the SSA is only about 0.92–0.93. These regions
ation is 0.04. In up to 75% of cases this parameter lies within the range
show peaks of smoke AOD, which is responsible for the absorption of
0.3–0.4, which is consistent with the vertical profiles of aerosol extinc­
solar radiation, especially during the summertime.
tion in the Arctic (Fig. 5).
Fig. 5 shows mean profiles of the aerosol extinction coefficient for
selected regions in the Arctic from NAAPS for the four seasons. The ABF
peak is in the lower troposphere; however, during spring the profiles are 3. Results
almost constant up to the upper troposphere. Dust particles are mostly
transported in the upper (summer) and middle troposphere. During 3.1. Mean ARF for the Arctic area
spring relatively high values of dust extinction coefficient are visible in
the entire troposphere. Smoke particles dominate during summer, while Table 5 shows the long-term annual mean of AOD, SW, LW, and net
the altitude for maximum extinction coefficients depends on the loca­ (SW + LW) ARF at the Earth’s surface and at TOA for different types of
tion. For example, over Ny-Alesund and Barrow the peaks are at about aerosols (anthropogenic and biogenic, mineral dust, smoke, sea salt),
total aerosol, aerosol at 5 km above the surface, and absorbing particles.

Fig. 4. Annual mean (a) visible (600 nm) surface albedo and (b) columnar aerosol single scattering albedo at 550 nm for the years 2015 and 2003–2015,
respectively.

8
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 5. Mean vertical profiles of aerosol extinction coefficient for ABF (a–d), dust (e–h), smoke (i–l), and sea salt particles (m–p) obtained from NAAPS reanalysis for
Ny-Alesund (blue lines), Barrow (red lines), Resolute Bay (green lines), and Ittoqqortoormiit (black lines). Four panels for each aerosol type correspond to winter
(December–February; a, e, i, m), spring (March–May; b, f, j, n), summer (June–August; c, g, k, o), and autumn (September–November; d, h, l, p), respectively.

9
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Different values were reported by Quinn et al. (2008) at the surface


(− 0.46 W/m2) and at TOA (0.30 W/m2) for anthropogenic particles as a
result of BC dominance in ARF (Breider et al., 2017). The mean TOA ARF
for the Arctic is comparable with a globally averaged value (− 0.27
W/m2) reported by Myhre et al. (2013), based on the AeroCom Phase II.
We found that particles above 5 km have a significant impact on ARF.
In the case of the surface ARF, it is about 22–23% (clear- and all-sky)
while for TOA ARF it is 14% (clear-sky). An even higher impact in the
Arctic was reported by Samset et al. (2013) for BC above 5 km (72%) due
to both a high fraction of aerosol at high altitudes and a strong forcing
efficiency caused by high surface albedo. For all-sky conditions, TOA
ARF by particles existing above 5 km is 0.2 W/m2, while ARF for total
aerosol is negative (− 0.1 W/m2). It can be explained by the appearance
of BC particles above the bright clouds, which enhances the solar ab­
sorption in the atmosphere (Feng et al., 2013).
In the previous section, we found that 12% of AOD is above high
clouds and 35% of AOD is above all clouds (Fig. 6). However, bearing in
mind that vertical profiles obtained from NAAPS reanalysis were not
validated thoroughly over the Arctic, similarly to other models (e.g.
Goddard Chemistry Aerosol Radiation and Transport and the Copernicus
Fig. 6. Probability density function of AOD fraction above the highest cloud Atmosphere Monitoring Service model). Nevertheless, Markowicz et al.
(red bars) and mean AOD fraction above clouds (blue bars). (2017a) indicated that NAAPS reanalysis was a reliable source for pre­
dicting intensive biomass burning events in the Arctic in July 2015 and
The latter category includes anthropogenic and biogenic, mineral dust, its vertical distribution of smoke matched qualitatively well with
and smoke aerosols. The data were averaged for the whole Arctic (north ground-based lidar observations. Recent observations of BC concentra­
of 70◦ N) under clear-sky and all-sky conditions. Total clear-sky SW ARF tion onboard tethered balloons in Ny-Alesund (Ferrero et al., 2016;
at the surface is − 4 W/m2, with contributions of 30%, 30%, 26%, and Markowicz et al., 2017b) provide some statistics on vertical variability
14% from anthropogenic and biogenic aerosol, mineral dust, smoke, and during spring and summer. However, such information is very limited in
sea salt particles, respectively. ARF by absorbing particles is around time and space and in vertical range, which usually reaches only 1–2 km.
− 3.4 W/m2, while including aerosol load above 5 km it exceeds − 0.9 W/ So the data is not used here, but it could be used for future boundary
m2. For the TOA, the cooling is much weaker (− 1.3 W/m2 for total layer aerosol validation once there is sufficient data. Recall that NAAPS
aerosol), and the contribution of aerosol types is different from at the reanalysis provides skilful AOD compared to the ground-based obser­
Earth’s surface. And it is 26%, 17%, 15%, and 43% for anthropogenic vations (Section 2.1). The uncertainties of SW ARF for total and
and biogenic aerosol, mineral dust, smoke, and sea salt, respectively. absorbing aerosol that mainly consists of smoke aerosols emitted from
The largest contribution to clear-sky TOA ARF is attributed to non- boreal fires and transported to the Arctic during summertime, are ex­
absorbing sea salt particles, due to the effect of scattering. pected to be low.
According to Table 5, cloud presence reduces surface cooling by In the case of LW ARF, the uncertainty, due mainly to uncertainties in
55–60%. In the case of TOA, clouds shift the ARF towards positive the optical parameters used in the calculation, can be very high, as
values. The mean SW all-sky TOA ARF for total aerosol is − 0.1 W/m2, indicated in the sensitivity experiments in Section 2.2. So the results of
while for anthropogenic and biogenic aerosol it is nearly 0 W/m2, and this study should be interpreted with caution. LW ARF at the surface is
for mineral dust, smoke, and sea salt it is 0.2, 0.05, and − 0.3 W/m2, about 25% of the absolute value of SW ARF. The surface LW ARF is
respectively. A similar study was performed by Breider et al. (2017) who higher than at TOA due to the existence of aerosol mostly in the lower
obtained an anthropogenic ARF of − 1.20 ± 0.05 W/m2 at the surface troposphere, where the temperature is close to the surface value.
and of − 0.19 ± 0.05 W/m2 at TOA, for long-term average over Therefore, the far-infrared effect is more important at the surface than at
2005–2010 and latitudes north of 60oN and all-sky conditions. Our data TOA (Markowicz et al., 2003). Net ARF, which is a sum of SW and LW, is
averaged for the same area show very similar results (− 1.05 W/m2 at − 3.0 W/m2 and -0.7 W/m2 for total aerosol at the surface and at TOA
surface and − 0.19 W/m2 at TOA). A slightly lower value (− 0.12 W/m2 under clear-sky conditions, respectively. For all-sky conditions, the
at TOA) was obtained by Sand et al. (2017) from AeroCom Phase II. respective values are − 1.9 and 0.3 W/m2. For anthropogenic and

Table 5
Annual mean AOD, clear-, and all-sky ARF [W/m2] for different aerosol types averaged for latitudes > 70.5◦ N.
Parameters AOD Sky conditions SW ARF [W/m2] LW ARF [W/m2] NET [W/m2]

SURF TOA SURF TOA SURF TOA

Anthropogenic 0.031 clear-sky − 1.29 − 0.36 0.49 0.43 − 0.80 0.07


all-sky − 0.79 − 0.01 0.26 0.26 − 0.53 0.25
Dust 0.017 clear-sky − 1.28 − 0.23 0.54 0.23 − 0.74 0.00
all-sky − 0.80 0.19 0.30 0.16 − 0.50 0.35
Smoke 0.013 clear-sky − 1.10 − 0.21 0.16 0.12 − 0.94 − 0.09
all-sky − 0.70 0.05 0.10 0.08 − 0.60 0.13
Sea Salt 0.026 clear-sky − 0.59 − 0.59 0.04 0.02 − 0.55 − 0.57
all-sky − 0.30 − 0.29 0.02 0.01 − 0.28 − 0.28
TOTAL 0.087 clear-sky − 4.00 − 1.34 0.99 0.61 − 3.01 − 0.73
all-sky − 2.42 − 0.09 0.54 0.39 − 1.88 0.31
Total aerosol above 5 km 0.016 clear-sky − 0.93 − 0.19 0.26 0.26 − 0.67 0.07
all-sky − 0.58 0.22 0.14 0.20 − 0.44 0.41
Absorbing 0.061 clear-sky − 3.36 − 0.63 1.15 0.73 − 2.21 0.10
all-sky − 2.10 0.29 0.62 0.47 − 1.48 0.76

10
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

biogenic aerosol, however, the all-sky ARF is − 0.5 and 0.3 W/m2 at the conditions, with the highest values observed over Greenland and lowest
surface and at TOA, respectively. The strongest surface cooling and over Northern Atlantic Ocean. Similar spatial patterns for anthropogenic
simultaneous TOA warming occur for absorbing particles (− 1.5 and 0.8 all-sky ARF were found by Sand et al. (2017). The maximum ARF re­
W/m2). The effect of mineral dust at TOA is neutral (0 W/m2) for ported over Greenland (up to 1 W/m2) is slightly higher than in this
clear-sky, and positive (0.4 W/m2) when clouds are included. Smoke study. Also, the negative ARF over northern Eurasia (up to − 1 W/m2)
particles at the surface show a cooling effect (− 0.6 W/m2) and at TOA obtained from AeroCom II is higher than those obtained in this study (up
are slightly warming (0.1 W/m2) for all-sky conditions. Finally, for sea to − 1.5 W/m2).
salt, all-sky ARF is the same at the surface and at TOA (− 0.3 W/m2) due
to negligible solar absorption. We found that aerosol higher than 5 km 3.3. Statistics of ARF over selected regions of the Arctic
above the surface has a significant cooling effect (− 0.4 W/m2) at the
surface, and warming (0.4 W/m2) at TOA, for all-sky conditions. For A set of demonstrative regions is chosen for further illustration of the
clear-sky conditions, its effect on the radiation budget is negative at the results the discussions below. They are “Greenland” (28◦ -55◦ W, 60◦ -
surface (− 0.7 W/m2) and positive at TOA (0.1 W/m2). 80◦ N), “Svalbard” (5◦ -30◦ E, 75◦ -83◦ N), and “Central Arctic” (85◦ -90◦ N).
These regions were selected in the European Arctic to show difference of
3.2. Spatial distribution of aerosol radiative forcing ARF due to surface reflectance and distance to middle latitudes. Table 6
shows annual mean AOD, SSA, and net ARF at the surface and at TOA for
Figs. 7 and 8 show the spatial distribution of annual mean ARFs by clear-sky and all-sky conditions for these regions. Among the three re­
total aerosol under clear- and all-sky conditions. The spatial variability gions, the most polluted is Svalbard, with mean AOD of 0.1 at 550 nm,
of ARFs is similar to AOD due to fact that ARF is in the first approxi­ and with sea salt as the most important contributor. Over Greenland,
mation linear function of AOD. Total aerosol SW ARF at the surface AOD is dominated by anthropogenic and biogenic aerosols and mineral
varies from about − 10 W/m2 over the northern Atlantic to about − 1 W/ dust, while over the Central Arctic region, the prevailing impactors are
m2 over Greenland and Central Arctic. The peak of SW ARF corresponds the anthropogenic and biogenic aerosols and sea salt particles. There are
to the region with the highest AOD (maximum of sea salt particles). The some differences in ARF between the regions. Over Greenland and the
Pearson correlation coefficient between mean AOD and SW ARF is Central Arctic region surface ARF is negative and TOA ARF is positive,
− 0.83 and − 0.81 at the surface and at TOA, respectively. The spatial except for the case of sea salt. In Svalbard, both ARFs are negative under
distribution of TOA SW ARF is similar (Fig. 7b), but the values are clear-sky and all-sky conditions, and TOA ARF is positive for mineral
shifted toward positive. Over the northern Atlantic, the ARF is about − 7 dust, the absorbing aerosol category, and aerosol above 5 km. For all-sky
W/m2, while over Greenland and Central Arctic it is about 1.5 W/m2. In conditions, anthropogenic and biogenic aerosol perturbs the net radia­
the case of LW radiation, the range of ARF is 0.4–2.2 W/m2 at surface tion budget at the surface and at TOA by about − 0.3 and 0.5 W/m2 over
and 0.3–1.3 W/m2 at TOA. The smallest radiation effect was obtained Greenland and the Central Arctic while by − 0.7 and 0 W/m2 F. For
over the northern Atlantic and Greenland, and the highest over northern absorbing aerosols, the all-sky ARFs are − 0.7 and 1.4 W/m2 for
Eurasia. The latter is in agreement with the occurrence of anthropogenic Greenland, − 1.7 and 0.1 W/m2 for Svalbard, and − 0.8 and 1.4 W/m2 at
and smoke aerosol. the surface and at TOA, respectively, for the Central Arctic region. The
The spatial variability of net ARF is similar to that of SW ARF because effect of smoke particles in the perturbation of the radiation budget is
LW ARF is relatively small (about 20–25% of SW ARF). Nevertheless, the significant, even though the contribution of biomass burning particles to
net ARF is shifted toward positive values, and varies from − 9 to − 0.5 W/ total AOD is the smallest for each region. All-sky ARF for particles above
m2 at the surface and from − 7 to 2 W/m2 at TOA. The positive ARF at 5 km is − 0.4 and 0.9 W/m2 in Greenland, − 0.5 and 0.2 on Svalbard, and
TOA was obtained for Greenland and the Central Arctic region, due to − 0.3 and 0.6 W/m2 in the Central Arctic region, at the surface and at
the high surface albedo. ARF for all-sky conditions shows significantly TOA, respectively. We found that the most significant differences be­
smaller absolute values (Fig. 8), which agrees with attenuated properties tween regions are observed at TOA ARF, where negative or positive
of clouds with respect to ARF. In the presented results this effect alters radiation effects rely on the surface albedo.
ARF by about 40% at the surface. Although the spatial distributions of
ARF for clear and all-sky conditions are similar, there are some differ­ 3.4. Annual variability of AOD and ARF for selected regions
ences. The strongest cooling effect was found not only over the northern
Atlantic Ocean but also over Central and Eastern Siberia and Alaska Annual variability of monthly mean AOD and its components for
under all-sky conditions (Fig. 8a). At TOA, SW ARF varies spatially be­ three selected regions and for the whole Arctic (>70◦ N) is shown in
tween − 2.4 and 0.6 W/m2. Over the Central Arctic region, the impact of Fig. 11. The time variability of total AOD (Fig. 11a) is a consequence of
aerosol on the radiation budget is almost zero. If we include the effect in the different cycles of aerosol components. For anthropogenic and
the LW band, the net ARF over the Central Arctic region is nearly zero at mineral dust particles (Fig. 11b and c), the peaks are observed during
the surface and a little higher at TOA (0.5–1 W/m2). spring, while for smoke (Fig. 11d), they occur in summer, and for sea salt
Similar plots to Figs. 7–8 but for anthropogenic and biogenic aerosol – during autumn and winter. The increase of AOD due to anthropogenic
ARF are shown in Fig. 9 (clear-sky) and Fig. 10 (all-sky). The spatial aerosol in spring is a phenomenon known as the Arctic haze (Shaw,
distribution is similar to the ARF obtained for all aerosols. However, the 1987; Quinn et al., 2007; Tomasi et al., 2007; Tomasi et al., 2012). It is
absolute values are significantly smaller in comparison to the previously related to aerosol transport from middle latitudes (Willis et al., 2018).
mentioned cases. Surface SW ARF for clear-sky conditions varies from Results of long-term (1963–2015) simulation by the Hybrid
− 3 W/m2 (part of Scandinavia and Russia, Northern Atlantic, and Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT; Stein
Barents Sea) to nearly 0 W/m2 over Greenland and the Central Arctic et al., 2015) (see supplement material, Fig. S1) show that global air
region. TOA SW ARF varies between − 1.5 W/m2 (part of Scandinavia pollution crossroads over Svalbard. The most common source region for
and Russia, Northern Atlantic, and Barents Sea) and 0.5 W/m2 (Southern 144 h back-trajectories ended over Ny-Alesund at 0.5 km is Northern
Greenland). For all-sky conditions, the SW ARF varies between − 1.5 and Asia (Siberia). In case of the back-trajectories ended at 2, 5, and 10 km
0 W/m2 at the Earth’s surface and between − 0.5 and 0.5 W/m2 at TOA, the air mass is mostly transported from Western Europe and Atlantic
with the same regions of extreme values. The spatial variability of LW Ocean, Eastern North America and Central North America, respectively.
ARF is significant only for clear-sky conditions. Simulation yields the Relative frequency of air mass transport from latitude less than 550 N is
highest values for the northern part of Eurasia, and a very small radia­ 6, 17, 34 and 31% for back-trajectories ended at 0.5, 2, 5, and 10 km,
tion impact over Greenland. TOA net ARF varies between − 1 and 0.8 W/ respectively. This data shows also that transport from middle latitudes at
m2 for clear-sky conditions and between − 0.5 and 0.9 W/m2 for all-sky all altitudes (Fig. S3) is more frequent during winter than summer.

11
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 7. Annual mean of (a, b) SW ARF, (c, d) LW ARF, (e, f) net ARF [W/m2] for clear-sky conditions. The panels on the left-hand side (a, c, e) correspond to surface
ARF and the panels on the right-hand side (b, d, f) to TOA ARF.

12
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 8. As in Fig. 7, but for all-sky conditions.

Transport from middle and low latitudes is also partially responsible 0.79). For almost every month, the anthropogenic and biogenic AOD
for the occurrence of mineral dust (Fig. 11c) in the Arctic. Therefore, the over Greenland is significantly lower in comparison to Svalbard and the
Pearson correlation coefficient between monthly mean anthropogenic Central Arctic region. The anthropogenic and biogenic aerosol contrib­
and biogenic AOD as well as mineral dust AOD is relatively high (r2 = utes about 25–40% to total AOD during springtime. The climatological

13
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 9. Annual mean of anthropogenic and biogenic (a, b) SW and (c, d) LW as well as (e, f) net ARF [W/m2] for clear-sky conditions. The panels on the left-hand side
(a, c, e) correspond to the surface ARF, and the panels on the right-hand side (b, d, f) to TOA ARF.

14
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 10. As in Fig. 9, but for all-sky conditions.

mean (2003–2015) smoke aerosol AOD reaches 0.03 in summer, indi­ contribution to total AOD during January and February. Over
cating a contribution of about 30–50% to total AOD. Greenland, sea salt is less important, but still contributing 40–50% to
During winter and autumn, sea salt is the dominant component, total AOD. Thus, the annual AOD cycle differs slightly between regions.
especially over Svalbard (Fig. 11e). This aerosol type has a 60–70% For example, the months for maximum AOD appearances are April and

15
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Table 6
Annual mean AOD as well as clear-sky and all-sky net ARF [W/m2] for different aerosol types averaged for Greenland, Svalbard, and the Central Arctic region.
Aerosol type Sky cond. Greenland Svalbard Central Arctic Region

AOD SSA AOD SSA AOD SSA ARF [W/m2]

SURF TOA SURF TOA SURF TOA

Anth Clear 0.019 0.93 − 0.35 0.49 0.029 0.94 − 1.22 − 0.48 0.030 0.94 − 0.32 0.50
All − 0.27 0.50 − 0.70 − 0.01 − 0.26 0.47
Dust Clear 0.013 0.88 − 0.33 0.45 0.014 0.88 − 1.03 − 0.45 0.017 0.88 − 0.39 0.41
All − 0.26 0.55 − 0.61 0.13 − 0.30 0.59
Smoke Clear 0.008 0.90 − 0.32 0.30 0.010 0.91 − 0.88 − 0.33 0.012 0.91 − 0.57 0.30
All − 0.24 0.33 − 0.51 − 0.02 − 0.38 0.35
Sea Salt Clear 0.016 1.00 − 0.25 − 0.26 0.043 1.00 − 1.12 − 1.13 0.024 1.00 − 0.18 − 0.22
All − 0.15 − 0.15 − 0.52 − 0.52 − 0.10 − 0.12
TOTAL Clear 0.056 0.95 − 1.33 0.84 0.097 0.97 − 4.37 − 2.66 0.082 0.95 − 1.44 0.88
All − 0.95 1.12 − 2.38 − 0.69 − 1.02 1.17
>5 km Clear 0.023 0.93 − 0.50 0.72 0.013 0.92 − 0.82 − 0.28 0.014 0.92 − 0.39 − 0.35
All − 0.39 0.87 − 0.49 0.20 − 0.28 0.55
Absor-bing Clear 0.040 0.92 − 0.91 1.25 0.053 0.92 − 2.81 − 1.05 0.058 0.92 − 1.11 1.23
All − 0.71 1.37 − 1.66 0.13 − 0.84 1.37

Fig. 11. Monthly mean of AOD at 550 nm by (a) total aerosol, (b) anthropogenic and biogenic, (c) mineral dust, (d) smoke, and (e) sea salt obtained from NAAPS
reanalysis for all the selected regions and the whole Arctic.

16
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

May for Greenland, February for Svalbard, and March and April for the SW ARF during late spring and summer is up to − 12 W/m2 over the
Central Arctic. NAAPS AOD annual variability shows good agreement Svalbard region, − 8 W/m2 over the Central Arctic region, and − 5 W/m2
with ground-based observations in contrast to most of the models used over Greenland. In the case of the anthropogenic and biogenic aerosols,
in the AeroCom II, which show a peak in AOD during summer (Sand the monthly mean forcing is up to − 4 W/m2, -2.5 W/m2, and -1.8 W/m2
et al., 2017). Only a few models from AeroCom II show a late spring over Svalbard, the Central Arctic region, and Greenland, respectively.
maximum. The LW ARF at the surface is shown in Fig. 13. In the case of the total
The annual cycle of clear-sky ARF is a combination of temporal aerosol, the annual cycle of the LW ARF is more closely correlated with
variability of diurnal TOA solar radiation (related to solar declination), that of AOD than of SW ARF. Thus, the maximum effect is obtained for
AOD, surface albedo, as well as single-scattering properties of aerosols. spring and the minimum during autumn (Fig. 13a). Also, for the
Fig. 12 shows monthly mean clear-sky SW ARF at the surface in the case anthropogenic and mineral dust aerosol the annual cycle is similar
of total aerosol (a), anthropogenic and biogenic aerosol (b), dust (c), (Fig. 13b). While the maximum ARF occurs during summer for smoke
smoke (d), and sea salt (e) for the three selected regions and for the (Fig. 12d), and winter for sea salt (Fig. 13e). The largest LW forcing is
whole Arctic. For all regions and almost all aerosol types, the maximum obtained from dust particles, due to their absorbing characteristics in the
radiation cooling at the surface is observed during summer, when LW. The maximum LW ARF for mineral particles is up to 1 W/m2 during
incoming TOA solar radiation is the highest. Only for the sea salt par­ spring. Although the uncertainty of the LW ARF is large, the annual cycle
ticles, the strongest cooling occurs during spring, because in summer sea is probably close to reality.
salt AOD is very low. Finally, we discuss the all-sky TOA SW ARF. In this case, the annual
The effect of sea salt on SW ARF is significantly higher over the variability is different, depending on the region and aerosol type
Svalbard region year around, except summertime, because it is sur­ (Fig. 14). For total aerosol, negative or zero ARF (Fig. 14a) occurs only
rounded by open water almost all year and it lies in a zone of high- over Svalbard throughout the year. Over Greenland, positive forcing
pressure gradient indicating enhanced production of sea salt. The total appears between May and September, while for the rest of the year it is

Fig. 12. Monthly mean of (a) total, (b) anthropogenic and biogenic, (c) mineral dust, (d) smoke, and (e) sea salt clear-sky SW ARF [W/m2] at the surface for
Greenland (navy blue), Svalbard (blue), the Central Arctic region (orange), and the whole Arctic (brown).

17
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 13. Monthly mean of (a) total, (b) anthropogenic and biogenic, (c) mineral dust, (d) smoke, and (e) sea salt clear-sky LW ARF [W/m2] at the surface for
Greenland (navy blue), Svalbard (blue), the Central Arctic region (orange), and the whole Arctic (brown).

almost zero. For the Central Arctic region, a significant positive ARF was 4. Summary and conclusions
computed for May, as well as negative ARF for March, and for the period
from June to September. Similar temporal variability is found for Perturbations of the radiation budget due to aerosols over the Arctic
anthropogenic and biogenic (Fig. 14b), mineral dust (Fig. 14c), and were discussed based on the 2003–2015 climatology of the aerosol op­
smoke aerosol (Fig. 14d). Only sea salt ARF is negative for all regions tical properties obtained from NAAPS reanalysis and radiative transfer
during the whole year, with the strongest cooling in spring and in simulation. The most important results are the following:
September (Fig. 14e). Annual variability of anthropogenic ARF reported
by Sand et al. (2017) shows also positive forcing in May for several - Annual AOD averaged over the Arctic (above 70.5◦ N) is 0.09 at 550
models and a negative peak for all models during summer (July, nm, while 0.09, 0.11, 0.09 and 0.06, for winter, spring, summer and
August). Similar temporal variation was found for smoke, which slightly autumn, respectively.
warms the system during spring and cools it during summer. Breider - NAAPS AOD contribution is 34% from anthropogenic and biogenic
et al. (2017) showed a significant change of anthropogenic ARF in the aerosol, 22% from mineral dust, 11% from smoke, and 33% from sea
periods 1980–1985 and 2005–2010, with emphasis on the decline in salt. AOD shows seasonal variability with a late autumn and winter
anthropogenic contribution to TOA ARF for spring due to the reduction peak for sea salt particles (53% of contribution to AOD), a spring
of sulphate particles at the surface (about 2–3%/year). If the trend maximum related to the Arctic haze phenomenon dominated by
continues in the future, one may expect very low negative ARF or anthropogenic aerosol (40%) and sea salt (22%), and a summer peak
slightly positive ARF during the Arctic haze period. related to the long-range transport of smoke aerosol (34%).
- Validation of model AOD versus AERONET data shows a relatively
small bias (0.01–0.02) for most stations and a relatively good
agreement (e.g. annual variability). Only over Greenland, AOD is

18
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Fig. 14. Monthly mean of (a) total, (b) anthropogenic and biogenic, (c) mineral dust, (d) smoke, and (e) sea salt all-sky SW ARF [W/m2] at TOA for Greenland (navy
blue), Svalbard (blue), the Central Arctic region (orange), and the whole Arctic (brown).

significantly overestimated similarly to other models (e.g. AeroCom SW ARF for anthropogenic and biogenic particles changes from
phase II, Sand et al., 2012) almost zero to negative values (− 0.2 W/m2) at TOA in all-sky con­
- Mean SW ARF for total aerosol shows strong cooling at the surface: 4 ditions. This result is similar to the value obtained during AeroCom
W/m2 for clear-sky and only − 1.3 W/m2 during all-sky conditions. Phase II (Sand et al., 2017).
At TOA, ARF is also negative for both cases (− 1.3 W/m2 and -0.4 W/ - Total SW ARF for the Arctic region is strongly correlated with surface
m2). albedo. Pearson correlation coefficient between surface albedo the
- Anthropogenic and biogenic particles have a cooling influence at the surface and TOA clear-sky SW ARF is 0.92 and 0.90, respectively. For
surface: 2.4 and − 0.8 W/m2 for clear-sky and all-sky conditions, AOD, the same quantity is 0.73, for TOA and 0.66 for surface ARF.
respectively. At TOA, ARF is slightly negative for clear-sky (− 0.1 W/ Negligible correlation coefficients are for SSA and surface (0.01) and
m2) and almost zero for all-sky conditions. TOA (0.12) ARF.
- For smoke and dust particles, the SW ARF is very similar both for the - The annual cycle of ARF in the Arctic is mainly a function of AOD,
Earth’s surface and for TOA, for clear-sky and all-sky conditions. surface albedo, and incoming solar radiation at TOA. As a result, the
Only for all-sky conditions at TOA, ARF is slightly positive (0.1–0.2 absolute value of ARF is large during both spring and summer times,
W/m2). because AOD is the highest in spring and TOA solar radiation is the
- For sea salt, SW ARF is the same at the surface and at TOA: 0.6 W/m2 maximum in summer.
for clear-sky and − 0.3 W/m2 for all-sky conditions, as a result of - The presence of cloud covers reduces the surface cooling (direct
negligible solar absorption. clear-sky SW ARF) by a factor of 40% and shifts TOA SW ARF to­
- Mostly due to the spatial variability of AOD and surface albedo the wards positive values.
mean SW ARF is sensitive to how the Arctic region is defined. If it is - Estimated LW ARF is significantly smaller than SW ARF. For dust as
considered as the region north of 60.5◦ N (expanding from 70.5oN), well as an anthropogenic and biogenic aerosol, LW ARF reaches 0.5

19
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

W/m2 at the surface and 0.15–0.3 W/m2 at TOA in all-sky conditions. Science Team and Atmospheric Composition Modeling and Analysis
Very low value (below 0.05 W/m2) has been found for sea salt TOA Program and the U.S. Office of Naval Research code 322.
and surface ARF.
- There are larger uncertainties in the LW ARF estimates than the SW Appendix A. Supplementary data
ARF estimates. The obtained values of net ARF are only an approx­
imation. A better estimation of LW ARF would require validations Supplementary data to this article can be found online at https://doi.
and potentially better representations of aerosol size distributions org/10.1016/j.atmosenv.2020.117882.
and aerosol vertical distributions in the model.
- The maximum uncertainties, by combing all maximum uncertainties
References
in the same direction in AFR calculation from a set of aerosol optical
properties, are between 26 and 29% for clear-sky and 29–38% for all- Abbatt, J.P.D., Leaitch, W.R., Aliabadi, A.A., Bertram, A.K., Blanchet, J.-P., Boivin-
sky conditions for SW AFR. While for LW ARF the numbers are 70%– Rioux, A., Bozem, H., Burkart, J., Chang, R.Y.W., Charette, J., Chaubey, J.P.,
100% for clear-sky and all-sky. Very high uncertainties in LW ARF is Christensen, R.J., Cirisan, A., Collins, D.B., Croft, B., Dionne, J., Evans, G.J.,
Fletcher, C.G., Galí, M., Ghahremaninezhad, R., Girard, E., Gong, W., Gosselin, M.,
due to the fact that extrapolating optical properties of aerosols from Gourdal, M., Hanna, S.J., Hayashida, H., Herber, A.B., Hesaraki, S., Hoor, P.,
visible and infrared to far-infrared is challenging and requires vali­ Huang, L., Hussherr, R., Irish, V.E., Keita, S.A., Kodros, J.K., Köllner, F., Kolonjari, F.,
dation of AOD at multiple longer wavelengths. Thus aerosol optical Kunkel, D., Ladino, L.A., Law, K., Levasseur, M., Libois, Q., Liggio, J., Lizotte, M.,
Macdonald, K.M., Mahmood, R., Martin, R.V., Mason, R.H., Miller, L.A.,
properties are much better constrained for SW than for LW. Also Moravek, A., Mortenson, E., Mungall, E.L., Murphy, J.G., Namazi, M., Norman, A.-L.,
since satellite retrieved AOD is a column integrated variable, aerosol O’Neill, N.T., Pierce, J.R., Russell, L.M., Schneider, J., Schulz, H., Sharma, S., Si, M.,
speciation and vertical distribution are not constrained by assimila­ Staebler, R.M., Steiner, N.S., Thomas, J.L., von Salzen, K., Wentzell, J.J.B., Willis, M.
D., Wentworth, G.R., Xu, J.-W., Yakobi-Hancock, J.D., 2019. Overview paper: new
tion of AOD. The operations used to convert AOD into 3-D speciated insights into aerosol and climate in the Arctic. Atmos. Chem. Phys. 19, 2527–2560.
aerosol fields and vice versa constitute another source of uncertainty https://doi.org/10.5194/acp-19-2527-2019.
in ARF estimation. Barry, R.G., 2017. The arctic cryosphere in the twenty-first century. Geogr. Rev. 107,
69–88. https://doi.org/10.1111/gere.12227.
- A large uncertainty of ARF in the LW can be reduce by improve
Bond, T.C., Streets, D.G., Yarber, K.F., Nelson, S.M., Woo, J.-H., Klimont, Z., 2004.
model representation of aerosol size distribution in the models, A technology-based global inventory of black and organic carbon emissions from
especially for coarse mode and their chemical composition. It can be combustion. J. Geophys. Res. 109, D14203. https://doi.org/10.1029/
2003JD003697.
done by validation of model’s size distribution by ground-based
Bond, Bond, T.C., Doherty, S.J., Fahey, D.W., Forster, P.M., Berntsen, T., DeAngelo, B.J.,
observation. In addition, the satellite retrieval of AOD in the Flanner, M.G., Ghan, S., Kärcher, B., Koch, D., Kinne, S., Kondo, Y., Quinn, P.K.,
middle-infrared can help to reduce the uncertainty of far-infrared Sarofim, M.C., Schultz, M.G., Schulz, M., Venkataraman, C., Zhang, H., Zhang, S.,
AOD estimation Bellouin, N., Guttikunda, S.K., Hopke, P.K., Jacobson, M.Z., Kaiser, J.W., Klimont, Z.,
Lohmann, U., Schwarz, J.P., Shindell, D., Storelvmo, T., Warren, S.G., Zender, C.S.,
2013. Bounding the role of black carbon in the climate system: a scientific
The rapid change of climate in the Arctic observed in the last de­ assessment. J. Geophys. Res. Atmos. 118, 5380–5552. https://doi.org/10.1002/
cades, with the increase in the surface temperature and the reduction of jgrd.50171.
Bourgeois, Q., Bey, I., 2011. Pollution transport efficiency toward the Arctic: sensitivity
the ice extent, has modified the aerosol impact on the Arctic system. The to aerosol scavenging and source regions. J. Geophys. Res. 116, D08213. https://doi.
reduction of surface albedo is moving the ARF towards negative values, org/10.1029/2010JD015096.
as is the increase of sea salt emission from ice-free regions (Struthers Breider, T.J., et al., 2017. Multi-decadal trends in aerosol radiative forcing over the
Arctic: contribution of changes in anthropogenic aerosol to Arctic warming since
et al., 2011). Therefore, the effect of aerosol on the Arctic amplification 1980. J. Geophys. Res. Atmos. 122 https://doi.org/10.1002/2016JD025321.
will probably decrease; however, detailed research is needed. Clarke, A.D., Noone, K.J., 1985. Soot in the Arctic snowpack: a cause for perturbations in
radiative transfer. Atmos. Environ. 19, 2045e2053.
Comiso, J.C., Parkinson, C.L., Gersten, R., et al., 2008. Accelerated decline in the Arctic
CRediT authorship contribution statement sea ice cover. Geophys. Res. Lett. 35, L01703.
Diehl, T., Heil, A., Chin, M., Pan, X., Streets, D., Schultz, M., Kinne, S., 2012.
K.M. Markowicz: Conceptualization, Methodology, Writing - orig­ Anthropogenic, biomass burning, and volcanic emissions of black carbon, organic
carbon, and SO2 from 1980 to 2010 for hindcast model experiments. Atmos. Chem.
inal draft. J. Lisok: Input to model preparation, Visualization, Writing -
Phys. Discuss. 12, 24895–24954. https://doi.org/10.5194/acpd-12-24895-2012.
review & editing. P. Xian: Formal analysis, Validation, NAAPS re- Dubovik, O., King, M.D., 2000. A flexible inversion algorithm for retrieval of aerosol
analysis preparation, model validation, Writing - review & editing. optical properties from Sun and sky radiance measurements. J. Geophys. Res. Atmos.
105, 20673–20696.
Dou, T.F., De Xiao, C., 2016. An overview of black carbon deposition and its radiative
Declaration of competing interest forcing over the Arctic. Adv. Clim. Change Res. 7 (3), 115–122.
Dubovik, O., Holben, B., Eck, T.F., Smirnov, A., Kaufman, Y.J., King, M.D., Tanré, D.,
The authors declare that they have no known competing financial Slutsker, I., 2002. Variability of absorption and optical properties of key aerosol
types observed in worldwide locations. J. Atmos. Sci. 59, 590–608.
interests or personal relationships that could have appeared to influence Dumont, M., Brun, E., Picard, G., et al., 2014. Contribution of light-absorbing impurities
the work reported in this paper. in snow to Greenland’s darkening since 2009. Nat. Geosci. 7 (7), 509–512.
Feng, Y., Ramanathan, V., Kotamarthi, V.R., 2013. Brown carbon: a significant
atmospheric absorber of solar radiation? Atmos. Chem. Phys. 13, 8607–8621.
Acknowledgements https://doi.org/10.5194/acp-13-8607-2013.
Ferrero, L., Cappelletti, D., Busetto, M., Mazzola, M., Lupi, A., Lanconelli, C., Becagli, S.,
The authors would like to acknowledge the support for this research Traversi, R., Caiazzo, L., Giardi, F., Moroni, B., Crocchianti, S., Fierz, M., Močnik, G.,
Sangiorgi, G., Perrone, M.G., Maturilli, M., Vitale, V., Udisti, R., Bolzacchini, E.,
from the Polish-Norwegian Research Programme operated by the Na­ 2016. Vertical profiles of aerosol and black carbon in the Arctic: a seasonal
tional Center for Research and Development under the Norwegian phenomenology along 2 years (2011–2012) of field campaigns. Atmos. Chem. Phys.
Financial Mechanism 2009–2014 within the framework of the Project 16, 12601–12629. https://doi.org/10.5194/acp-16-12601-2016.
Flanner, M.G., Zender, C.S., Randerson, J.T., et al., 2007. Present-day climate forcing and
Contract No. Pol-Nor/196911/38/2013. The authors also acknowledge response from black carbon in snow. J. Geophys. Res. Atmos. 112, D11202. https://
Brent Holben and Piotr Sobolewski for the use of the data from the doi.org/10.1029/2006JD008003.
AERONET station in Hornsund, Victoria E. Cachorro Revilla and Sandra Fu, Q., Liou, K.N., 1992. On the correlated k-distribution method for radiative transfer in
nonhomogeneous atmospheres. J. Atmos. Sci. 49 (22), 2139–2156.
Blindheim for the data from the Andenes AERONET station, Rick
Fu, Q., Liou, K.N., 1993. Parameterization of the radiative properties of cirrus clouds.
Wagener for Ittoqqortoormiit, Ihab Abboud and Vitali Fioletov for J. Atmos. Sci. 50 (13), 2008–2025.
Resolute Bay, and for Barrow AERONET station data. The data from Ny- Fu, Q., Thorsen, T.J., Su, J., Ge, J.M., Huang, J.P., 2009. Test of Mie-based single-
Ålesund sun-photometer were provided by the Alfred Wegener Institute scattering properties of non-spherical dust aerosols in radiative flux calculations.
J. Quant. Spectrosc. Radiat. Transf. 110 (14–16), 1640–1653.
for Polar and Marine Research. Dr. Xian’s Arctic NAAPS reanalysis work Gong, S.L., Zhao, T.L., Sharma, S., Toom-Sauntry, D., Lavoué, D., Zhang, X.B., 716
was sponsored by NASA Grant NNH17AE58I on behalf of the Aura Leaitch, W.R., Barrie, L.A., 2010. Identification of trends and interannual 717

20
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

variability of sulfate and black carbon in the Canadian High Arctic: 1981–2007,. Myhre, G., Kvalevag, M., Rädel, G., Cook, J., Shine, K.P., Clark, H., Karcher, F.,
J. 718 Geophys. Res. Atmospheres 115, D07305. https://doi.org/10.1029/ Markowicz, K., Kardas, A., Wolkenberg, P., Balkanski, Y., Ponater, M., Forster, P.,
2009JD012943. Rap, A., Rodriguez De Leon, R., et al., 2009. Intercomparison of radiative forcing
Granier, C., Bessagnet, B., Bond, T., D’Angiola, A., van der Gon, H.D., Frost, G.J., Heil, A., calculations of stratospheric water vapour and contrails. Meteorol. Z. 18 (6),
Kaiser, J.W., Kinne, S., Klimont, Z., Kloster, S., Lamarque, J.-F., Liousse, C., pp585–596.
Masui, T., Meleux, F., Mieville, A., Ohara, T., Raut, J.-C., Riahi, K., Schultz, M.G., Myhre, G., Samset, B.H., Schulz, M., Balkanski, Y., Bauer, S., Berntsen, T.K., Bian, H.,
Smith, S.J., Thompson, A., van Aardenne, J., van der Werf, G.R., van Vuuren, D.P., Bellouin, N., Chin, M., Diehl, T., Easter, R.C., Feichter, J., Ghan, S.J.,
2011. Evolution of anthropogenic and biomass burning emissions of air pollutants at Hauglustaine, D., Iversen, T., Kinne, S., Kirkevåg, A., Lamarque, J.-F., Lin, G., Liu, X.,
global and regional scales during the 1980–2010 period. Climate Change 109, Lund, M.T., Luo, G., Ma, X., van Noije, T., Penner, J.E., Rasch, P.J., Ruiz, A.,
163–190. Seland, Ø., Skeie, R.B., Stier, P., Takemura, T., Tsigaridis, K., Wang, P., Wang, Z.,
Hänel, G., 1976. The properties of atmospheric aerosol particles as functions of relative Xu, L., Yu, H., Yu, F., Yoon, J.-H., Zhang, K., Zhang, H., Zhou, C., 2013. Radiative
humidity at thermodynamic equilibrium with surrounding moist air. Adv. Geophys. forcing of the direct aerosol effect from AeroCom Phase II simulations. Atmos. Chem.
19, 73–188, 1976. Phys. 13, 1853–1877. https://doi.org/10.5194/acp-13-1853-2013.
Hegg, D.A., Covert, D.S., Crahan, K., Jonsson, H.H., 2002. The dependence of aerosol Overland, J.E., Wang, M., 2013. When will the summer arctic be nearly sea ice free?
light-scattering on RH over the Pacific Ocean. Geophys. Res. Lett. 29 https://doi. Geophys. Res. Lett. 40 (10), 2097e2101. https://doi.org/10.1002/grl.50316.
org/10.1029/2001GL014495, 60-61–60-4. Platnick, S., King, M.D., Ackerman, S.A., Menzel, W.P., Baum, B.A., Riedi, J.C., Frey, R.
Hess, M., Koepke, P., Schult, I., 1998. Optical properties of aerosols and clous: the A., 2003. The MODIS cloud products: algorithms and examples from Terra. IEEE
software package OPAC. Bulletin of the Americal Meteorologic Soctiety 79 (5), Trans. Geosci. Rem. Sens. 41, 459–473.
831–844. Polyakov, I.V., Pnyushkov, A.V., Alkire, M., Ashik, I.M., Baumann, T., Carmack, E.C.,
Hirdman, D., Sodemann, H., Eckhardt, S., Burkhart, J.F., Jefferson, A., Mefford, T., Goszczko, I., Guthrie, J., Ivanov, V.V., Kanzow, T., Krishfield, R., Kwok, R.,
Quinn, P.K., Sharma, S., Ström, J., Stohl, A., 2010. Source identification of short- Sundfjord, A., Morison, J., Rember, R., Yulin, A., 2017. Greater role for atlantic
lived air pollutants in the Arctic using statistical analysis of measurement data and inflows on sea-ice loss in the eurasian basin of the arctic ocean. Science. https://doi.
particle dispersion model output. Atmos. Chem. Phys. 10, 669-693. org/10.1126/science.aai8204.
Hogan, T., Rosmond, T., 1991. The description of the navy operational global Quinn, P.K., Shaw, G., Andrews, E., Dutton, E., Ruoho-Airola, T., Gong, S., 2007. Arctic
atmospheric prediction system’s spectral forecast model. Mon. Weather Rev. 119, Haze: current trends and knowledge gaps. Tellus 59B, 99–114.
1786–1815. Quinn, P.K., et al., 2008. Short-lived pollutants in the Arctic: their climate impact and
Hyer, E.J., Reid, J.S., Zhang, J., 2011. An over-land aerosol optical depth dataset for data 812 possible mitigation strategies. Atmos. Chem. Phys. 8 (6), 1723–1735. https://
assimilation by filtering, correction, and aggregation of MODIS Collection 5 optical doi.org/10.5194/acp-8-1723-2008, 813.
depth retrievals. Atmos. Meas. Tech. 4, 379–408. https://doi.org/10.5194/amt-4- Reid, J.S., Koppmann, R., Eck, T.F., Eleuterio, D.P., 2005a. A review of biomass burning
379-2011. emissions part II: intensive physical properties of biomass burning particles. Atmos.
Intergovernmental Panel on Climate Change, 2018. Summary for policymakers. In: Chem. Phys. 5, 799–825. https://doi.org/10.5194/acp-5-799-2005.
Masson-Delmotte, V., Zhai, P., Pörtner, H.O., Roberts, D., Skea, J., Shukla, P.R., et al. Reid, J.S., Eck, T.F., Christopher, S.A., Koppmann, R., Dubovik, O., Eleuterio, D.P.,
(Eds.), Global warming of 1.5 ◦ c. An IPCC special report on the impacts of global Holben, B.N., Reid, E.A., Zhang, J., 2005b. A review of biomass burning emissions
warming of 1.5 ◦ c above pre-industrial levels and related global greenhouse gas part III: intensive optical properties of biomass burning particles. Atmos. Chem.
emission pathways, in the context of strengthening the global response to the threat Phys. 5, 827–849. https://doi.org/10.5194/acp-5-827-2005.
of climate change, sustainable development, and efforts to eradicate poverty. World Reid, J.S., Hyer, E.J., Prins, E.M., Westphal, D.L., Zhang, J., Wang, J., Christopher, S.A.,
Meteorological Organization, Geneva, Switzerland, p. 32. Curtis, C.A., Schmidt, C.C., Eleuterio, D.P., Richardson, K.A., Hoffman, J.P., 2009.
Jaeglé, L., Quinn, P.K., Bates, T.S., Alexander, B., Lin, J.-T., 2011. Global distribution of Global monitoring and forecasting of biomass-burning smoke: description of and
sea salt aerosol: new constraints from in situ and remote sensing observations. lessons from the fire locating and modeling of burning emissions (FLAMBE)
Atmos. Chem. Phys. 11, 3137–3157. https://doi.org/10.5194/acp-11-3137-2011. Program. IEEE J. Sel. Top. Appl. 2, 144–162. JSTARS-2009-00034.
Koch, D., Del Genio, A.D., 2010. Black carbon semi-direct effects on cloud cover: review Ren, L., Yang, Y., Wang, H., Zhang, R., Wang, P., Liao, H., 2020. Source attribution of
and synthesis. Atmos. Chem. Phys. 10, 7685–7696. https://doi.org/10.5194/acp-10- Arctic aerosols and associated Arctic warming trend during 1980–2018. Atmos.
7685-2010. Chem. Phys. Discuss. https://doi.org/10.5194/acp-2020-3 (submitted for
Koepke, P., Gasteiger, J., Hess, M., 2015. Technical Note: Optical properties of desert publication).
aerosol with non-spherical mineral particles: data incorporated to OPAC. Atmos. Saltzman, E.S., King, D.B., Holmen, K., Leck, C., 1993. Experimental determination of the
Chem. Phys. 15, 5947–5956. https://doi.org/10.5194/acp-15-5947-2015.3. diffusion coefficient of dimethylsulfide in water. J. Geophys. Res. 98, 16481–16486.
Kok, J.F., Ridley, D.A., Zhou, Q., Miller, R.L., Zhao, C., Heald, C.L., Ward, D.S., Albani, S., Samset, B.H., Myhre, G., Schulz, M., Balkanski, Y., Bauer, S., Berntsen, T.K., Bian, H.,
Haustein, K., 2017. Smaller desert dust cooling effect estimated from analysis of dust Bellouin, N., Diehl, T., Easter, R.C., Ghan, S.J., Iversen, T., Kinne, S., Kirkevag, A.,
size and abundance. Nat. Geosci. 10, 274–278. https://doi.org/10.1038/ngeo2912. Lamarque, J.-F., Lin, G., Liu, X., Penner, J.E., Seland, Ø., Skeie, R.B., Stier, P.,
Lana, A., Bell, T.G., Simó, R., Vallina, S.M., Ballabrera-Poy, J., Kettle, A.J., Dachs, J., Takemura, T., Tsigaridis, K., Zhang, K., 2013. Black carbon vertical profiles strongly
Bopp, L., Saltzman, E.S., Stefels, J., Johnson, J.E., Liss, P.S., 2011. An updated affect its radiative forcing uncertainty. Atmos. Chem. Phys. 13, 2423–2434. https://
climatology of surface dimethlysulfide concentrations and emission fluxes in the doi.org/10.5194/acp-13-2423-2013.
global ocean. Global Biogeochem. Cycles 25, GB1004. https://doi.org/10.1029/ Sand, M., Berntsen, T.K., Kay, J.E., Lamarque, J.F., Seland, Ø., Kirkevag, A., 2013. The
2010GB003850. Arctic response to remote and local forcing of black carbon. Atmos. Chem. Phys. 13,
Lewis, E.R., Schwartz, S.E., 2004. Sea Salt Aerosol Production: Mechanisms, Methods, 211–224. https://doi.org/10.5194/acp-13-211-2013.
Measurements, and Models: A Critical Review. American Geophysical Union, Sand, M., Samset, B.H., Balkanski, Y., Bauer, S., Bellouin, N., Berntsen, T.K., Bian, H.,
Washington, DC, 2004. Chin, M., Diehl, T., Easter, R., Ghan, S.J., Iversen, T., Kirkevåg, A., Lamarque, J.-F.,
Li, D., Shine, K.P., 1995. In: A 4 Dimensional Ozone Climatology for UGAMP Models, Lin, G., Liu, X., Luo, G., Myhre, G., van Noije, T., Penner, J.E., Schulz, M., Seland, Ø.,
UGAMP Internal Rep.. Dept. of Meteorol., Cent. for Global and Atmos. Modell., Univ. Skeie, R.B., Stier, P., Takemura, T., Tsigaridis, K., Yu, F., Zhang, K., Zhang, H., 2017.
of Reading, Reading, U.K, 35, 35. Aerosols at the Poles: an AeroCom phase II multi-model evaluation. Atmos. Chem.
Li-Jones, X., Maring, H.B., Prospero, J.M., 1998. Effect of relative humidity on light Phys. Discuss. https://doi.org/10.5194/acp-2016-1120 (submitted for publication).
scattering by mineral dust aerosol as measured in the marine boundary layer over Sessions, W.R., Reid, J.S., Benedetti, A., Colarco, P.R., da Silva, A., Lu, S., Sekiyama, T.,
the tropical Atlantic Ocean. J. Geophys. Res. 103, 31113–31121. Tanaka, T.Y., Baldasano, J.M., Basart, S., Brooks, M.E., Eck, T.F., Iredell, M.,
Lynch, P., Reid, J.S., Westphal, D.L., Zhang, J., Hogan, T.F., Hyer, E.J., Curtis, C.A., Hansen, J.A., Jorba, O.C., Juang, H.-M.H., Lynch, P., Morcrette, J.-J., Moorthi, S.,
Hegg, D.A., Shi, Y., Campbell, J.R., Rubin, J.I., Sessions, W.R., Turk, F.J., Walker, A. Mulcahy, J., Pradhan, Y., Razinger, M., Sampson, C.B., Wang, J., Westphal, D.L.,
L., 2016. An 11-year global gridded aerosol optical thickness reanalysis (v1.0) for 2015. Development towards a global operational aerosol consensus: basic
atmospheric and climate sciences. Geosci. Model Dev. (GMD) 9, 1489–1522. climatological characteristics of the international cooperative for aerosol prediction
Markowicz, K.M., Witek, M., 2011. Sensitivity study of global contrail radiative forcing multi-model ensemble (ICAP-MME). Atmos. Chem. Phys. 15, 335–362.
due to particle shape. J. Geophys. Res. 116, D23203. https://doi.org/10.1029/ Sharma, S., Ishizawa, M., Chan, D., Lavoué, D., Andrews, E., Eleftheriadis, K.,
2011JD016345. Maksyutov, S., 2013. 16-year simulation of Arctic black carbon: transport, source
Markowicz, K.M., Flatau, P.J., Vogelmann, A.M., Quinn, P.K., Welton, E.J., 2003. Clear- contribution, and sensitivity analysis on deposition. J. Geophys. Res. Atmos. 118,
sky infrared aerosol radiative forcing at the surface and the top of the atmosphere. 943–964. https://doi.org/10.1029/2012JD017774.
Q. J. R. Meteorol. Soc. 129, 2927–2947. https://doi.org/10.1256/qj.02.224. Shaw, G., 1987. The Arctic haze phenomenon. Bull. Am Met. Soc 76, 2403–2413.
Markowicz, K.M., Lisok, J., Xian, P., 2017a. Simulations of the effect of intensive biomass Shi, Y., Zhang, J., Reid, J.S., Liu, B., Hyer, E.J., 2014. Critical evaluation of cloud
burning in July 2015 on Arctic radiative budget. Atmos. Environ. 171, 248–260. contamination in the MISR aerosol products using MODIS cloud mask products.
Markowicz, K.M., Ritter, C., Lisok, J., Makuch, P., Stachlewska, I.S., Cappelletti, D., Atmos. Meas. Tech. 7, 1791–1801. https://doi.org/10.5194/amt-7-1791-2014,
Mazzola, M., Chilinski, M.T., 2017b. Vertical variability of aerosol single-scattering 2014.
albedo and black carbon concentration based on in-situ and remote sensing Shindell, D., Faluvegi, G., 2009. Climate response to regional radiative forcing during the
techniques during iAREA campaigns in Ny-Ålesund, Atmospheric Environment, 164, twentieth century. Nat. Geosci. 2, 294–300.
431–447. Sofiev, M., Soares, J., Prank, M., de Leeuw, G., Kukkonen, J., 2011. A regional-to-global
Ming, Y., Russell, L.M., 2001. Predicted hygroscopic growth of sea salt aerosol. model of emission and transport of sea salt particles in the atmosphere. J. Geophys.
J. Geophys. Res. Atmos. 106, 28259–28274. Res. 116, D21302. https://doi.org/10.1029/2010JD014713.
Monahan, E.C., Spiel, D.E., Davidson, K.L., 1986. A model of marine aerosol generation Stein, A.F., Draxler, R.R., Rolph, G.D., Stunder, B.J.B., Cohen, M.D., Ngan, F., 2015.
via whitecaps and wave disruption. In: Monahan, E.C., MacNiocaill, G. (Eds.), NOAA’s HYSPLIT atmospheric transport and dispersion modeling system. Bull. Am.
Oceanic Whitecaps and Their Role in Air-Sea Exchange Processes, vols. 167–174. Meteorol. Soc. 96, 2059–2077.
Springer, New York.

21
K.M. Markowicz et al. Atmospheric Environment 244 (2021) 117882

Struthers, H., Ekman, A.M.L., Glantz, P., Iversen, T., Kirkevåg, A., Mårtensson, E.M., Witek, M.L., Flatau, P.J., Quinn, P.K., Westphal, D.L., 2007b. Global sea-salt modeling:
Seland, Ø., Nilsson, E.D., 2011. The effect of sea ice loss on sea salt aerosol results and validation against multicampaign shipboard measurements. J. Geophys.
concentrations and the radiative balance in the Arctic. Atmos. Chem. Phys. 11, Res. 112, D08215. https://doi.org/10.1029/2006JD007779.
3459–3477. https://doi.org/10.5194/acp-11-3459-2011. Witek, M.L., Flatau, P.J., Teixeira, J., Markowicz, K.M., 2011. Numerical investigation of
Teixeira, J., Hogan, T.F., 2002. Boundary layer clouds in a global atmospheric model: sea salt aerosol size bin partitioning in global transport models: implications for mass
simple cloud cover parameterizations. J. Clim. 15 (11), 1261–1276. budget and optical depth. Aerosol Sci. Technol. 45 (3), 401–414.
Tomasi, C., et al., 2007. Aerosols in polar regions, 2007, A historical overview based on Witek, M.L., Diner, D.J., Garay, M.J., 2016. Satellite assessment of sea spray aerosol
optical depth and in situ observations. J. Geophys. Res. 112, D16205. https://doi. productivity: southern Ocean case study. J. Geophys. Res. Atmos. 121, 872–894.
org/10.1029/2007JD008432. https://doi.org/10.1002/2015JD023726.
Tomasi, C., et al., 2012. An update on polar aerosol optical properties using POLAR-AOD Xian, P., Reid, J.S., Hyer, E.J., et al., 2019. Current state of the global operational aerosol
and other measurements performed during the International Polar Year, 52. Atmos. multi-model ensemble: an update from the International Cooperative for Aerosol
Environ. 52, 29–47. Prediction (ICAP). Q. J. R. Meteorol. Soc. 145 (Suppl. 1), 176–209. https://doi.org/
Twomey, S., 1977. The influence of pollution on the shortwave albedo of clouds. 10.1002/qj.3497, 2019.
J. Atmos. Sci. 34 (7), 1149–1152. Yang, Q., Bitz, C.M., Doherty, S.J., 2014. Offsetting effects of aerosols on Arctic and
Westphal, D.L., Toon, O.B., Carlson, T.N., 1988. A case study of mobilization and global climate in the late 20th century. Atmos. Chem. Phys. 14 (8), 3969–3975.
transport of Saharan dust. J. Atmos. Sci. 45, 2145–2175. https://doi.org/10.5194/acp-14-3969-2014, 919.
Willis, M.D., Leaitch, W.R., Abbatt, J.P., 2018. Processes controlling the composition and Zhang, J., Reid, J.S., 2006. MODIS aerosol product analysis for data assimilation:
abundance of Arctic aerosol. Rev. Geophys. 56, 621–671. https://doi.org/10.1029/ assessment of over-ocean level 2 aerosol optical thickness retrievals. J. Geophys. Res.
2018RG000602. 111, D22207. https://doi.org/10.1029/2005JD006898.
Witek, M.L., Flatau, P.J., Teixeira, J., Westphal, D.L., 2007a. Coupling an ocean wave Zhang, J., Reid, J.S., Westphal, D.L., Baker, N.L., Hyer, E.J., 2008. A system for
model with a global aerosol transport model: a sea salt aerosol parameterization operational aerosol optical depth data assimilation over global oceans. J. Geophys.
perspective. Geophys. Res. Lett. 34, L14806. https://doi.org/10.1029/ Res. 113, D10208. https://doi.org/10.1029/2007JD009065.
2007GL030106.

22

You might also like