TCThe CryosphereTCThe Cryosphere1994-0424Copernicus PublicationsGöttingen, Germany10.5194/tc-10-477-2016The darkening of the Greenland ice sheet: trends, drivers, and projections (1981–2100)TedescoMarcocryocity@gmail.comDohertySarahFettweisXavierhttps://orcid.org/0000-0002-4140-3813AlexanderPatrickJeyaratnamJeyavinothStroeveJulienneLamont-Doherty Earth Observatory of the Columbia University, New York, Palisades, NY, USANASA Goddard Institute of Space Studies, New York, NY, USAThe City College of New York – CUNY, New York, NY, USAUniversity of Liege, Liege, BelgiumNASA Goddard Institute for Space Studies, New York, NY, USAThe Graduate Center of the City University of New York, New York, NY, USAUniversity of Boulder, Boulder, CO, USAMarco Tedesco (cryocity@gmail.com)3March20161024774966September201519October201513January201610February2016This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://tc.copernicus.org/articles/10/477/2016/tc-10-477-2016.htmlThe full text article is available as a PDF file from https://tc.copernicus.org/articles/10/477/2016/tc-10-477-2016.pdf
The surface energy balance and meltwater production of the Greenland ice
sheet (GrIS) are modulated by snow and ice albedo through the amount of
absorbed solar radiation. Here we show, using space-borne multispectral data
collected during the 3 decades from 1981 to 2012, that summertime
surface albedo over the GrIS decreased at a statistically significant (99 %)
rate of 0.02 decade-1 between 1996 and 2012. Over the same
period, albedo modelled by the Modèle Atmosphérique Régionale (MAR)
also shows a decrease, though at a lower rate (∼-0.01 decade-1)
than that obtained from space-borne data. We suggest that the
discrepancy between modelled and measured albedo trends can be explained by
the absence in the model of processes associated with the presence of
light-absorbing impurities. The negative trend in observed albedo is
confined to the regions of the GrIS that undergo melting in summer, with the
dry-snow zone showing no trend. The period 1981–1996 also showed no
statistically significant trend over the whole GrIS. Analysis of MAR outputs
indicates that the observed albedo decrease is attributable to the combined
effects of increased near-surface air temperatures, which enhanced melt and
promoted growth in snow grain size and the expansion of bare ice areas, and
to trends in light-absorbing impurities (LAI) on the snow and ice surfaces.
Neither aerosol models nor in situ and remote sensing observations indicate
increasing trends in LAI in the atmosphere over Greenland. Similarly, an
analysis of the number of fires and BC emissions from fires points to the
absence of trends for such quantities. This suggests that the apparent
increase of LAI in snow and ice might be related to the exposure of a “dark
band” of dirty ice and to increased consolidation of LAI at the surface with
melt, not to increased aerosol deposition. Albedo projections through to the
end of the century under different warming scenarios consistently point to
continued darkening, with albedo anomalies averaged over the whole ice sheet
lower by 0.08 in 2100 than in 2000, driven solely by a warming climate.
Future darkening is likely underestimated because of known underestimates in
modelled melting (as seen in hindcasts) and because the model albedo scheme
does not currently include the effects of LAI, which have a positive
feedback on albedo decline through increased melting, grain growth, and darkening.
Introduction
The summer season over the Greenland ice sheet (GrIS) during the past 2
decades has been characterized by increased surface melting (Nghiem et al.,
2012; Tedesco et al., 2011, 2014) and net mass loss (Shepherd et al., 2012).
Notably, the summer of 2012 set new records for surface melt extent (Nghiem
et al., 2012) and duration (Tedesco et al., 2013), and a record of
570 ± 100 Gt in total mass loss, doubling the average annual loss rate of
260 ± 100 Gt for the period 2003–2012 (Tedesco et al., 2014).
Net solar radiation is the most significant driver of summer surface melt
over the GrIS (van den Broeke et al., 2011; Tedesco et al., 2011), and is
determined by the combination of the amount of incoming solar radiation and
surface albedo. Variations in snow albedo are driven principally by changes
in snow grain size and by the presence of light-absorbing impurities (LAI,
Warren and Wiscombe, 1980). Generally, snow albedo is highest immediately
following new snowfall. In the normal course of destructive metamorphism, the snow grains become
rounded, and large grains grow at the expense of small grains, so the
average grain radius r increases with time (LaChapelle, 1969). Subsequently,
warming and melt/freeze cycles catalyse grain growth, decreasing albedo
mostly in the near-infrared (NIR) region (Warren, 1982). The absorbed solar
radiation associated with this albedo reduction promotes additional grain
growth, further reducing albedo, potentially accelerating melting. The
presence of LAI such as soot (black carbon, BC), dust, organic matter, algae,
and other biological material in snow or ice also reduces the albedo, mostly
in the visible and ultraviolet regions (Warren, 1982). Such impurities are
deposited through dry and wet deposition, and their mixing ratios are
enhanced through snow water loss in sublimation and melting (Conway et al.,
1996; Flanner et al., 2007; Doherty et al., 2013). Besides grain growth and
LAI, another cause of albedo reduction over the GrIS is the exposure of bare
ice: once layers of snow or firn are removed through ablation, the exposure
of the underlying bare ice will further reduce surface albedo, as does the
presence of melt pools on the ice surface (e.g. Tedesco et al., 2011).
Most of the studies examining albedo over the whole GrIS have focused on
data collected by the Moderate Resolution Imaging Spectroradiometer (MODIS)
starting in 2000 (e.g. Box et al., 2012; Tedesco et al., 2013). At the same
time, regional climate models (RCMs) have been employed to simulate the
evolution and trends of surface quantities over the GrIS back to the 1960s
using reanalysis data for forcing (e.g. Fettweis et al., 2013). Despite the
increased complexity of models, and their inclusion of increasingly
sophisticated physics parameterizations, RCMs still suffer from incomplete
representation of processes that drive snow albedo changes, such as the
spatial and temporal distribution of LAI, and from the absence of in situ
grain size measurement to validate modelled snow grain size evolution. In
this study, we first report the results from an analysis of summer albedo
over the whole GrIS from satellite for the period 1980–2012, hence
expanding the temporal coverage with respect to previous studies. Then, we
combine the outputs of an RCM and in situ observations with the satellite
albedo estimates to identify those processes responsible for the observed
albedo trends. The model, Modèle Atmosphérique Régionale (MAR),
is used to simulate surface temperature, grain size, exposed ice area, and
surface albedo over Greenland at large spatial scales. MAR-simulated surface
albedo is tested against surface albedo retrieved under the Global LAnd
Surface Satellite (GLASS) project, and it is used to attribute trends in
GLASS albedo. Lastly, we project the evolution of mean summer albedo over
Greenland using the MAR model forced with the outputs of different Earth
system models (ESMs) under different CO2 scenarios. Discussion and
conclusions follow the presentation of the methods and results.
Methods and dataThe MAR regional climate model and its albedo scheme
Simulations of surface energy balance quantities over the GrIS are performed
using the Modèle Atmosphérique Régionale (MAR; e.g. Fettweis et
al., 2005, 2013). MAR is a modular atmospheric model that uses the
sigma-vertical coordinate to simulate airflow over complex terrain and the
Soil Ice Snow Vegetation Atmosphere Transfer scheme (SISVAT, e.g. De Ridder
and Gallée, 1998) as the surface model. MAR outputs have been assessed
over Greenland in several studies (e.g. Tedesco et al., 2011; Fettweis et
al., 2005; Vernon et al., 2013; Rae et al., 2012; van Angelen et al., 2012),
with recent work specifically focusing on assessing simulated albedo over
Greenland (Alexander et al., 2014). A discussion of this evaluation is
presented later in the manuscript. The snow model in MAR is the CROCUS model
of Brun et al. (1992), which calculates albedo for snow and ice as a
function of snow grain properties, which in turn are dependent on energy and
mass fluxes within the snowpack. The model configuration used here has
25 terrain-following sigma layers between the Earth's surface and the
5 hPa model top. The spatial configuration of the model uses the 25 km
horizontal resolution computational domain over Greenland described in
Fettweis et al. (2005). The lateral and lower boundary conditions are
prescribed from meteorological fields modelled by the global European Centre
for Medium-Range Weather Forecasts (ECMWF) Interim Reanalysis (ERA-Interim,
http://www.ecmwf.int/en/research/climate-reanalysis/era-interim).
Sea-surface temperature and sea-ice cover are also prescribed in the model
using the same reanalysis data. The atmospheric model within MAR interacts
with the CROCUS model, which provides the state of the snowpack and
associated quantities (e.g. albedo, grain size). No nudging or interactive
nesting was used in any of the experiments.
The MAR albedo scheme is summarized below. Surface albedo is expressed as a
function of the optical properties of snow, the presence of bare ice,
whether snow is overlying ice (and whether the surface is waterlogged), and
the presence of clouds. In the version used here (MARv 3.5.1), the broadband
albedo (αs, 0.3–2.8 µm) of snow is a weighted average
(Eq. 1) of the albedo in three spectral bands, α1, α2,
and α3, which are functions of the optical diameter of
snow grains (d, in metres), as modified from equations by Brun et al. (1992;
e.g. Lefebre et al., 2003; Alexander et al., 2014):
αs=0.58α1+0.32α2+0.10α3α1=max(0.94,0.96-1.58d),(0.3-0.8µm)α2=0.95-15.4d,(0.8-1.5µm)α3=364⋅min(d,0.0023)-32.31d+0.88,(1.5-2.8µm).
The optical diameter d is, in turn, a function of snow grain properties and
it evolves as described in Brun et al. (1992). In MAR, the albedo of snow
is calculated by Eqs. (1)–(4), but it is not permitted to drop below 0.65.
For the transition from snow to ice, MAR makes the albedo an explicit
function of density. On a polar ice sheet, densification of snow/firn/ice
occurs in three stages, with a different physical process responsible for
the densification in each stage (Herron and Langway Jr., 1980; Arnaud et al.,
2000). Newly fallen snow can have density in the range 50–200 kg m-3.
After then, densification can occur due to wind processes, which break and
round grains, forming windslab of density typically around 300–400 kg m-3.
The remaining densification happens by grain-boundary sliding, attaining a maximum
density of ∼ 550 kg m-3 at the surface. Old melting snow
at the surface in late summer typically has this density, but does not
exceed it, because this is the maximum density that can be attained by
grain-boundary sliding and corresponds to the density of random-packing of
spheres (Benson, 1962, p. 77). Further increases of density (the second
stage) occur in firn under the weight of overlying snow, by grain deformation
(pressure-sintering). In this case the density range is 550–830 kg m-3.
At a density of 830 kg m-3 the air becomes closed off into bubbles and
the material is called ice. In the third stage, the density of ice increases
from 830 to 917 kg m-3 by shrinkage of air bubbles under pressure.
Moving down the slope along the surface of the GrIS, at the transition
between the accumulation area and the ablation area, the snow melts away,
exposing firn. Continuing farther down, the firn melts away, exposing ice.
The albedo of firn may be approximated as a function of its density, ρ,
interpolating between the minimum albedo of snow and the maximum albedo
of ice. In MAR these values of albedo are set to 0.65 and 0.55,
respectively. We would then have for the density range of firn (550–830 kg m-3)
αfirn=0.55+(0.65-0.55)(830-ρ)/(830-550).
The MARv3.5.1 version used here maintains a minimum albedo of 0.65 for any
density up to 830 kg m-3, and specifies the gradual transition from
snow albedo to ice albedo across the density range 830–920 kg m-3. This
means that the albedo of exposed firn is not allowed to drop below 0.65,
with the result that the positive feedbacks of snow/firn/ice albedo will be
muted in MAR. This aspect is being addressed in future versions of MAR (MAR v3.6)
and a sensitivity analysis is being conducted to evaluate the impact
of the changes on the albedo values when snow is transitioning from firn to
ice. Such analysis is computationally expensive and preliminary outputs will
be published once available.
In MAR, the albedo for bare ice is a function of the accumulated surface
meltwater preceding runoff and specified minimum (αi,min) and
maximum (αi,max) bare ice values:
αi=αi,min+αi,max-αi,mine(-MSW(t)K).
Here αi,min and αi,max are set, respectively, to 0.4
and 0.55, K is a scale factor set to 200 kg m-2, and MSW(t) is
the time-dependent accumulated excess surface meltwater before runoff (in kg m-2).
When a snowpack with depth less than 10 cm is overlying a layer with a
density exceeding 830 kg m-3 (i.e. ice), the albedo in MAR is a
weighted, vertically averaged value of snow albedo
(αS) and ice albedo (αI; e.g. if snow
depth is 3 cm then albedo is obtained by multiplying the snow albedo by 0.3
and adding the ice albedo multiplied by 0.7). When the snowpack depth
exceeds 10 cm, the value is set to αS. The presence of clouds
can increase snow albedo because they absorb at the same NIR wavelengths
where snow also absorbs, skewing the incident solar spectrum to wavelengths
for which snow has higher albedo (Fig. 5 of Grenfell et al., 1981; Fig. 13 of
Warren, 1982; Greuell and Konzelman, 1994), in which case the albedos
of snow and ice are adjusted based on the cloud fraction modelled by MAR
(Greuell and Konzelman, 1994).
The GLASS albedo product
The GLASS surface albedo product (http://glcf.umd.edu/data/abd/) is derived
from a combination of data collected by the Advanced Very High Resolution
Radiometer (AVHRR) and the Moderate Resolution Imaging Spectroradiometer
(MODIS, Liang et al., 2013). Shortwave broadband albedo (0.3–3 µm) is
provided every 8 days at a spatial resolution of 0.05∘ (∼ 56 km
in latitude) for the period 1981–2012. GLASS
albedo data with a resolution of 1 km are also available from 2000 to 2012
but these data are not used here for consistency with the data available before 2000.
There have been several efforts to make the AVHRR and MODIS albedo products
consistent within the GLASS product, including the use of the same surface
albedo spectra to train the regression and the use of a temporal filter and
climatological background data to fill data gaps (Liang et al., 2013).
Monthly averaged broadband albedos from GLASS-AVHRR and GLASS-MODIS were
cross-compared over Greenland for those months when there was overlap (July 2000,
2003, and 2004), revealing consistency in GLASS-retrieved albedo from
the two sensors (He et al., 2013). More information on the GLASS data
processing algorithm and product is available in Zhao et al. (2013) and Ying
et al. (2014).
The GLASS product provides both black-sky albedo (i.e. albedo in the
absence of a diffuse component of the incident radiation) and white-sky
albedo (albedo in the absence of a direct component, with an isotropic
diffuse component). The actual albedo is a value interpolated between these
two according to the fraction of diffuse sunlight, which is a function of
the aerosol optical depth (AOD) and cloud cover fraction. In the absence of
the full information needed to properly reconstruct the actual albedo, here
we use in our analysis the black-sky albedo, because we focus mostly on
albedo retrieved under clear-sky conditions. Our analysis using the
white-sky albedo (not shown here) is fully consistent with the results
obtained using the black-sky albedo and reported in the following. A full
description of the GLASS retrieval process and available products can be
found in Liang et al. (2013) and references therein. An assessment of the
GLASS product complementing existing studies is reported below.
Data collected by the MODIS Terra and Aqua sensors are used in the GLASS
albedo retrieval for the period 2000–2012 (2000–2012 for Terra and
2002–2012 for Aqua, respectively). Wang et al. (2012) have shown that the
MODIS Terra sensor has been degrading at a pace that can be approximated by
a second-order polynomial, with the coefficients being spectrally dependent.
Over Greenland, the impact of sensor degradation on albedo trends has been
estimated at -0.0059 decade-1 (Stroeve et al., 2013). Polashenski et
al. (2015) found a much greater impact on retrieved broadband albedo from Terra
sensor degradation (-0.03 decade-1). However, Polashenski et al. (2015) use
a daily product (MOD10A1) rather than a 16-day integrated product as in the
case of GLASS (e.g. Ying et al., 2014), which does account for the bidirectional reflectance distribution function (BRDF) at high
solar zenith angles. The performance of the MODIS daily product has been
shown to deteriorate with latitude (e.g. Alexander et al., 2015). On the
other hand, the use of the BRDF (as in the case of the GLASS product)
improves the performance of the product at high latitudes (Alexander et al.,
2015). This, together with the good agreement between the MCD43 albedo
product and the surface station albedo data (Alexander et al., 2014), gives
us confidence in the GLASS trends.
We complement previous assessments of the MODIS and GLASS albedo, evaluating
the absolute accuracy of the GLASS retrievals by comparing monthly GLASS
albedo to in situ measurements of albedo collected at automatic weather
stations of the Greenland climate network (GC-Net, Steffen and Box, 2001).
GC-Net data are distributed at hourly temporal resolution and were
temporally averaged to match the temporal window used in the GLASS product
data. The root mean square error (RMSE), percentage RMSE (RMSEp), and the
slope of a linear fit between GLASS and in situ measured albedos for
12 stations are given in Table 1. The number of available years used for the
statistics is also reported for each station. We considered only stations
for which at least 10 years were available for the analysis in at least one
of the months. Our results are consistent with the findings reported by
Alexander et al. (2014) and Stroeve et al. (2005, 2013) concerning the
assessment of the MODIS albedo products over the GrIS. The mean value of the
RMSE for all stations is 0.04–0.05 in all months, with individual station
values as high as 0.15 for station JAR1 in August and as low as 0.01 for
Summit and Saddle stations in June. The relatively large RMSE value for JAR1
(and other stations located within the ablation zone) is probably due to the
heterogeneity of albedo values within the pixel containing the location of
the station and to the point-scale nature of the in situ observations. At
Summit, where spatial inhomogeneity on the surface is small, it is
reasonable to assume that the effect of the spatial scale and heterogeneity on
the comparison is smaller.
ResultsAlbedo trends
The time series of the mean summer GLASS albedo values between 1981 and 2012
over Greenland can be separated into two distinct periods
(Fig. 1a): the period 1981–1996, when albedo
shows no trend, and a second period, 1996–2012, when a statistically
significant trend (99 %) is detected. The year 1996 was identified as
yielding the highest value of the coefficient of determination when fitting
the albedo time series with two linear functions using a variable breaking point.
Mean summer standardized values plotted as time series for
(a) albedo from GLASS (black) and MAR (grey), together with MAR-simulated values
of (b) surface air temperature, (c) surface grain size (effective radius of
optically “equivalent” sphere), and (d) bare ice exposed area. Trends for
the periods 1981–1996 and 1996–2012 are reported in each plot. Trends
in (a) refer to the GLASS albedo. The baseline 1981–2012 period is used
to compute standardized anomalies, obtained by subtracting the mean and then
dividing by the standard deviation of the values in the time series. All
trends are computed from JJA averaged values over ice-covered areas only,
not tundra.
Comparison between GLASS-retrieved albedo and GC-NET in situ albedo
measurements, for monthly averaged and seasonally averaged albedos at 12 surface
stations on the Greenland ice sheet.
June July August JJA StationRMSERMSEpSlopeNo. ofRMSERMSEpSlopeNo. ofRMSERMSEpSlopeNo. ofRMSERMSEpSlopeNo. of(%)yearsyearsyearsyearsSwiss0.1219.60-0.22110.023.861.1290.046.921.0080.022.731.067CP0.078.720.12120.067.400.14140.067.210.11130.078.20-0.0211Humboldt0.0810.38-0.1680.079.310.3590.089.980.39100.079.420.278Summit0.011.450.85150.022.25-0.25160.011.71-0.68160.011.220.1215TunuN0.056.72-0.66150.067.890.79150.078.840.69150.067.530.3715Dye-20.022.580.57140.022.150.75140.011.730.68150.011.540.8212Jar10.068.450.68130.1023.800.68150.1543.550.22140.0714.240.6612Saddle0.011.280.94140.021.950.61140.011.750.46140.011.310.7114NASAE0.034.230.46140.055.970.14140.045.110.24140.044.970.2414NASA SE0.022.760.59130.022.320.67130.022.140.36140.022.230.5613JAR20.0612.270.20110.0510.00-0.10120.0611.96-0.06110.048.510.1610Mean0.0487.130.04556.990.059.20.0385.62
Maps of JJA trends (per decade) from 1996 to 2012, when darkening
began to occur, for (a) space-borne-estimated GLASS albedo, (b) number of days
when MAR-simulated surface air temperature exceeded 0 ∘C,
(c) MAR-simulated surface grain size, and (d) number of days when bare ice is
exposed as simulated by MAR. Regions where trends are not significant at a
95 % level are shown as grey-hatched areas. White regions over the north
end of the ice sheet indicate areas which were not viewed by the satellite.
Differences between space-borne-measured and model-simulated albedo
trends in different spectral regions. (a) Difference between the GLASS and
MAR trends (albedo change per decade), with positive values indicating those
regions where MAR trend is smaller in magnitude than GLASS. Maps of JJA mean
albedo trends (1996–2012) simulated by MAR for (b) visible and
(c) near-infrared wavelengths.
The GLASS albedo shows significant darkening (p< 0.01) of the
surface of the GrIS for the 1996–2012 period, with the summer (JJA)
albedo declining at a rate of 0.02 ± 0.004 decade-1 (Fig. 1a). About 25 % of this decline might
be attributed to sensor degradation, per the analysis of Stroeve et al. (2013). However, the Terra sensor degradation is spectrally dependant and
temporally non-linear (Wang et al., 2012). This, together with the fact that
the GLASS product uses a combination of both Terra and Aqua data (which
reduces the impact of the Terra sensor degradation) indicates that the impact of
the sensor degradation on the observed decline is much smaller than 25 %.
Over the same period, MAR-simulated summer near-surface temperature
increased at a rate of 0.74 ± 0.5 ∘C decade-1
(Fig. 1b, p< 0.05), consistent with
observed enhanced surface melting (e.g. Fettweis et al., 2013). MAR
simulations also point to positive trends between 1996 and 2012 in summer
surface grain radius (0.12 ± 0.03 mm decade-1, p< 0.01,
Fig. 1c) and the extent of those regions where
bare ice is exposed during summer (380 ± 190 km2 decade-1,
p< 0.01, Fig. 1d). There is no
statistically significant trend in GLASS summer albedo or MAR-simulated
surface grain size and bare ice extent for the 1981–1996 period. Simulated
summer snowfall (not plotted in the figure) does not show a statistically
significant trend for the period 1996–2012 (p< 0.1, -1702 ± 790 mmWE decade-1).
Notably, strong negative summer snowfall anomalies
from 2010 to 2012 are simulated by MAR, down to -1.5 standard deviations
below the 1981–2012 mean. We suggest that for 2010–2012, in addition to
surface melting, reduced summer snowfall might have played a key role in the
accelerated decline in summer albedo.
Drivers: surface grain size and bare ice
Interannual variability in the mean summer GLASS albedo is captured by the
MAR albedo simulations (Fig. 1a). For the period
when the darkening has been identified, MAR albedo values explain
∼ 90 % (de-trended) of the space-borne-derived summer albedo
interannual variability. A multilinear regression analysis indicates that,
over the same period, the interannual variability of summer values of
surface grain size and bare ice extent simulated by MAR explain,
respectively, 54 % (grain size) and 65 % (bare ice) of the
interannual variability of GLASS albedo when considered separately. When
linearly combined, grain size, bare ice extent, and snowfall explain
∼ 85 % of the GLASS interannual variability, with the
influence of summer new snowfall alone explaining only 44 % of the GLASS
summer albedo variability.
The spatial distribution of observed summer albedo trends from space shows
that the largest trends (in magnitude) occur over those regions where
surface temperature, grain size, and bare ice exposure have also changed the
most (Fig. 2). In particular, darkening observed
from space is most pronounced at lower elevations in southwest Greenland,
with trends as large as -0.20 ± 0.07 decade-1
(Fig. 2a; note that the colour bar only goes down to -0.06 decade-1 for graphical purposes),
where trends in the number of days when simulated surface temperature exceeds 0 ∘C
(Fig. 2b), grain size (Fig. 2c), and the number of summer days when bare ice is exposed
(Fig. 2d) are the largest.
(a) MAR and (b) GLASS mean JJA albedo for the year 2010 over an area
including the dark band together with (c) time series of mean JJA albedo for
the ice-covered areas in the black rectangle. The black line in (c) shows the
GLASS spatially averaged albedo, where the top and the bottom of the grey
area indicate, respectively, the maximum and minimum albedo within the black
box in (b).
While MAR is able to capture a large component of the observed variability
in albedo retrieved by GLASS, the simulated albedo trend is smaller in
magnitude than that estimated using the GLASS product. The largest
differences occur along the southwest margin of the ice sheet
(Fig. 3), where a “dark band” of outcropping
layers of ice containing large concentrations of LAI is known to be present
on the surface (Wientjes et al., 2011). In this region the number of days
when surface temperature exceeds 0 ∘C has increased, with
trends of up to more than 20 days decade-1 along the margins of the
GrIS (Fig. 1b). During this time period, GLASS
albedo values are as low as 0.30, lower than that of bare ice (i.e. 0.45),
consistent with in situ measured values of dirty ice (Wientjes and Oerlemans, 2010;
Bøggild et al., 2010). Figure 4 shows the
spatial distribution of MAR and GLASS mean JJA albedo for the year 2010 over an
area centred on the dark band in southwest Greenland, as well as the time
series of GLASS albedo averaged over the same ice-covered area contained
within the region identified by the black rectangle in
Fig. 4a. The black line in Fig. 4c shows the GLASS spatially averaged albedo
within this region, with the top and the bottom margins of the grey area indicating,
respectively, the maximum and minimum albedo values within that area. Note that we
included only pixels that contained 100 % ice in all years (i.e. coloured areas
in Fig. 4a and b) in the calculation shown in Fig. 4c, so trends are not
driven by exposure of underlying land surface. Mean summer albedo from GLASS
decreased over this area between 2005 and 2012 from ∼ 0.6 to ∼ 0.45
(vs. a decrease simulated by MAR of 0.075). Minimum summer
albedo across all years averaged over the region is ∼ 0.4, but
dips close to ∼ 0.3 in 2010, a value consistent with dirty bare
ice, as shown in previous studies (Wientjes and Oerlemans, 2010; Wientjes et al., 2011;
Bøggild et al., 2010). We hypothesize that the discrepancy along this dark
band between MAR and GLASS albedo values is likely due to trends in the
concentrations of LAI in the snow and ice in this region, which are not currently captured by the model.
Time series of modelled and measured mean summer (JJA) albedo
anomalies (with respect to the year 2000) in different spectral bands.
(a) Visible, near-infrared, and shortwave-infrared albedo values simulated by
MAR; (b) as in (a) but for the visible albedo only from MAR, MODIS (obtained
from the product MCD43), and GLASS. Note that the vertical axis scale in (b)
is different from that in (a).
Maps of MAR-simulated albedo trends between 1996 and 2012 using
(a) the original MAR albedo scheme and (b) the perturbated MAR outputs in which
daily albedo is artificially decreased by 0.1 from the MAR-computed value
for those regions where bare ice is exposed. (c) Difference between the
trends obtained with MAR original albedo scheme and the perturbated solution.
Drivers: light-absorbing impurities on the surface of the GrIS
MAR simulations of albedo in different spectral bands (see Eqs. 1–4) point to
comparable trends in the visible (0.3–0.8 µm; -0.009 ± 0.005 decade-1,
p< 0.05) and near-infrared (0.8–1.5 µm;
-0.010 ± 0.004 decade-1, p< 0.05) bands
(Fig. 5a) and to a much smaller trend in the shortwave-infrared band that is not statistically significant (1.5–2.8 µm, -0.003 ± 0.004 decade-1,
p> 0.1). Because the GLASS product does not provide visible albedo
(only broadband albedo), we extrapolated an estimate of the visible component
of the GLASS albedo by subtracting the NIR and shortwave-infrared albedo values
computed with MAR from the GLASS broadband values, following the MAR albedo
scheme (Eq. 1, Fig. 5b). To evaluate the robustness of this approach, we
compared anomalies (with respect to the year 2000) in estimated GLASS visible albedo
with those from the 16-day MODIS MCD43A3 product (Stroeve et al., 2013), which
also has a visible albedo product (Fig. 5b). The MODIS albedo product we
used is distributed by Boston University (https://lpdaac.usgs.gov/)
and makes use of all atmospherically corrected MODIS reflectance measurements
over 16-day periods, to provide an averaged albedo every 8 days. A semi-empirical
bidirectional reflectance distribution function (BRDF) model is used to compute
bi-hemispherical reflectance from these reflectance measurements (Schaaf et al.,
2002). The comparison between the GLASS- and MODIS-retrieved visible albedo
anomalies is shown in Fig. 5b, indicating that the two visible albedo anomalies
are highly consistent, with a mean absolute error of 0.01 and a standard deviation
of 0.005. There are differences in the estimated summer albedo trends from MCD43A3
and GLASS over the 2002–2012 period, with the former being -0.04 ± 0.001 decade-1
and the latter -0.03 ± 0.008 decade-1. This difference could be due
to the method we applied to estimate the visible component of the GLASS albedo,
as well as other factors related to the data processing and algorithms used to
extract albedo. Notably, however, the GLASS and MCD43 visible albedo trends are
consistently about twice that estimated from the MAR model. The underestimated
darkening by MAR relative to GLASS can be attributed to several factors, including
the modelled spatial and temporal variability of the exposed bare ice area and
the concentration of surface LAI on the ice surface, which is currently not
included in the MAR albedo scheme. A lack of impurities in the MAR albedo scheme
can affect simulated albedo trends in at least two ways: first, the concentration
of impurities over bare ice areas could be increasing, which would not be
captured by MAR; second, the lack of impurities in the MAR albedo scheme causes
bare ice areas to have an overestimated albedo. More frequent exposure of bare
ice would lead to a decline in annual average albedo over time, but if the
underlying bare ice is darker, such a trend would be larger. Thus, the difference
in trends could result solely from an overestimation of the bare ice albedo by
MAR. We are not able to discern the degree to which the difference is due to
(a) errors in the area and frequency of bare ice exposure from MAR, (b) increasing
concentration of impurities not captured by MAR, or (c) overestimation of albedo
of an unchanging impurity-covered bare ice surface. The study by Alexander et
al. (2015) suggests that bare ice albedo is, indeed, overestimated in MAR. To
test the impact of a fixed bare ice albedo on the simulated albedo trend, we
performed a sensitivity experiment in which daily albedo for those pixels showing
bare ice exposure is reduced by a fixed value of 0.1. The magnitude of the
difference in trends between the original MAR simulation (with no change on the
bare ice albedo) and the one with a modified albedo (Fig. 6) being comparable to
the difference between the MAR and GLASS trends (Fig. 3a) suggests that this
factor alone could explain the difference. To further investigate this aspect,
we test the hypothesis of increased concentration of LAI on the snow and ice
surface. The concentrations of LAI in surface snow and ice can increase either
because of increased atmospheric deposition or because of post-depositional
processes, including (a) loss of snow water to sublimation and melt, resulting
in impurities accumulating at the surface as a lag deposit (e.g. Doherty et
al., 2013), and (b) the outcropping of “dirty” underlying ice associated with
snow/firn removal due to ablation. These processes are themselves driven by
warming, and therefore constitute positive feedbacks.
Quantifying the contribution of surface LAI to GLASS summer albedo trends is
a challenging task because of the relatively low impurity concentrations
over most of the GrIS (Doherty et al., 2010; Bond et al., 2013), and because
of known limitations related to remote sensing estimates of LAI from space
(Warren, 2013). Moreover, quantifying the causes of potential increased
impurity concentrations on the surface (atmospheric deposition vs. other
factors) is also challenging, if not prohibitive, given the current
state-of-the-art space-borne measurements (e.g. accuracy of the satellite
products) and the scarcity of in situ data. Therefore, in the next section,
we look for trends in forest fires and the emissions of BC from forest fires
in the main source regions for aerosols over the GrIS and assess whether
atmospheric aerosol concentrations over the GrIS have increased (as a proxy
for whether the deposition of aerosol has increased).
Attribution: aerosol contributions to LAI in GrISTrends in GrIS LAI
Ice core analyses of black carbon in the central regions of the GrIS have
been used to study long-term variability and trends in pollution deposition
(McConnell et al., 2007; Keegan et al., 2014). These records show that snow
at these locations was significantly more polluted in the first half of the
twentieth century than presently. Both these records and in situ
measurements at Summit (Cachier and Pertuisot, 1994; Chýlek et al., 1995;
Hagler et al., 2007; Doherty et al., 2010) also indicate that in recent
decades, the snow in central Greenland has been relatively clean, with
concentrations smaller than 4 ng g-1 for BC. This amount of BC could
lower snow albedo by only 0.002 for r= 100 µm, or 0.005 for
r= 500 µm (Fig. 5a of Dang et al., 2015). More recently,
Polashenski et al. (2015) analysed BC and dust concentrations in 2012–2014 snowfall along a
transect in northwest Greenland. They found similarly low concentrations of
BC and concluded that albedo decreases in their study region are unlikely to
be attributable to increases in BC or dust. Black carbon measurements from a
high snowfall region of west central Greenland made on an ice core collected
in 2003 show that black carbon concentrations varied significantly during
the previous 215 years, with an average annual concentration of 2.3 ng g-1
during the period 1952–2002, characterized by high year-to-year
variability in summer and a gradual decline in winter BC concentrations
through the end of the century (McConnell et al., 2007). Snow sampled in
1983 at Dye-3 had a median of 2 ng g-1 (Clarke and Noone, 1985). In
2008 and 2010, measurements 160 km away at Dye-2, using the same method, had
medians of 4 ng g-1 in spring and 1 ng g-1 in summer (Table 9 of
Doherty et al., 2010).
In the absence of in situ measurements of impurity concentration trends over
Greenland more broadly, or of trends in aerosol deposition rates (which are
absent entirely), we investigate trends in emissions from key sources of
aerosols deposited to the GrIS and trends in atmospheric aerosol optical depth (AOD)
over GrIS.
Trends in fire count and BC emissions
Biomass burning in North America and Siberia is a significant source of
combustion aerosol (BC and associated organics) to the GrIS (Hegg et al.,
2009, 2010). Therefore, we investigated trends in the number of active fires
in these two source regions, as well as BC emissions from fires in
subregions within the Northern Hemisphere. For fire counts we used the
MODIS monthly active fire products produced by the Terra (MOD14CMH) and Aqua
sensors (MYD14CMH) generated at 0.5∘ spatial resolution and
distributed by the University of Maryland via anonymous ftp (http://www.fao.org/fileadmin/templates/gfims/docs/MODIS_Fire_Users_Guide_2.4.pdf,
http://modis.gsfc.nasa.gov/data/dataprod/dataproducts.php?MOD_NUMBER=_14). The results of our analysis are summarized in
Fig. 7, showing the standardized (subtracting the mean and dividing by the
standard deviation of the 2002–2012 baseline period) cumulative number of
fires (April through to August) detected over North America (NA) and Eurasia (EU)
by the MOD14CMH and MYD14CMH GCM climatology products between 2002 and
2012. The figure shows large interannual variability but no significant
trend (at 90 % level) in the number of fires over the two areas between
2002 and 2012. The period between 2004 and 2011, when enhanced melting
occurred over the GrIS, shows a negative trend (though also in this case not
statistically significant).
In addition to number of fires we looked for trends specifically in BC
emissions from fires in potential source regions for GrIS, using estimates
from the Global Fire Emissions Database (GFED version 4.1,
http://www.globalfiredata.org/). There is a great deal of interannual
variability in annual BC emissions from fires in all regions (Fig. 8),
with no statistically significant increase during the 1997–2012 or 1997–2014
periods from either of the Boreal source regions or from central Asia or
Europe. BC emissions from fires in temperate North America increased by, on
average, 0.35 × 109 g yr-1 during 1997–2014 and by
0.52 × 109 g yr-1 during 1997–2012 (p< 0.1 in both cases). However, the total BC emissions from fires in this region
constitute a small fraction of that from the Boreal regions. In addition,
the only statistically significant trend in regional BC emissions is a
decrease in central Asia (112.6 × 109 g yr-1; p- 0.02), when GrIS albedo has
declined most precipitously. Xing et al. (2013, 2015) point out that direct
anthropogenic emissions have also been decreasing across almost all of the
mid- to high-latitude Northern Hemisphere.
Standardized cumulative number of fires (April through to August)
detected over North America (NA) and Eurasia (EU) by the MOD14CMH and
MYD14CMH GCM climatology products between 2002 and 2012.
June–July–August mean and standard deviation of measured aerosol
optical depth (AOD) at 550 nm at the three sites of Thule, Ittoqqortoormiit,
and Kangerlussuaq of the AERONET network (AERONET website, http://aeronet.gsfc.nasa.gov, 2013).
NA = not available.
Station YearThuleIttoqqortoormiitKangerlussuaq77∘28′00′′ N, 69∘13′50′′ W70∘29′07′′ N, 21∘58′00′′ W67∘00′31′′ N, 50∘41′21′′ W20070.042 ± 0.010NANA20080.040 ± 0.017NA0.051 ± 0.01220090.093 ± 0.020NA0.088 ± 0.01720100.052 ± 0.0110.052 ± 0.0050.049 ± 0.00720110.060 ± 0.0170.072 ± 0.0410.053 ± 0.01220120.065 ± 0.0110.044 ± 0.0090.072 ± 0.02020130.050 ± 0.0070.053 ± 0.0090.066 ± 0.010Trends in AOD over Greenland
To investigate trends in AOD over GrIS we look at AOD as simulated by models
and as measured at ground-based stations at several locations around the
GrIS. AOD is a measure of the total extinction (omnidirectional scattering
plus absorption) of sunlight as it passes through the atmosphere, and is
related to atmospheric aerosol abundance. Thus, it is a metric for the mass
of aerosol available to be potentially deposited onto the GrIS surface. In
the aerosol models, we are able to examine trends in total AOD as well as in
aerosol components: BC, dust, and organic matter. In addition, we examined
trends in modelled deposition fluxes of these species to the GrIS.
BC emissions (g) from fires in potential source regions for GrIS
for (a) all fire types and (b) boreal fires only using estimates from the
Global Fire Emissions Database (GFED version 4.1, http://www.globalfiredata.org/)
between 1997 and 2014.
NASA GISS ModelE standardized deposition fluxes for BC, dust, and
organic aerosol (OA) at Kangerlussuaq for (a) June, (b) July, and
(c) August (1981–2008) from the AEROCOM simulations.
For our analysis, we used model results from the Aerosol Comparisons between
Observations and Models (AeroCom) project, an open international initiative
aimed at understanding the global aerosol and its impact on climate (Samset
et al., 2014; Myhre et al., 2013; Jiao et al., 2014; Tsigaridis et al.,
2014). The project combines a large number of observations and outputs from
fourteen global models to test, document, and compare state-of-the-art
modelling of the global aerosol. We specifically show standardized (i.e. subtracting the mean and then dividing by the standard deviation) deposition
fluxes of BC, dust, and organic aerosol (OA) from the GISS modelE
contribution to the AeroCom phase II series of model runs
(http://aerocom.met.no/aerocomhome.html). The runs used here took as input
the decadal emission data from the Coupled Model Intercomparison Project
Phase 5 (CMIP5). In this case, we report the outputs of the NASA GISS ModelE
obtained from the AeroCom. In particular, Figs. 9 and 10 show modelled
deposition fluxes at the two locations of Kangerlussuaq (Fig. 9,
67∘00′31′′ N, 50∘41′21′′ W) and Summit
(Fig. 10, 72∘34′47′′ N, 38∘27′33′′ W) for the months of June, July, and August and aerosol
components (BC, dust, and organic matter). These locations were selected as
representative of the ablation zone (Kangerlussuaq) and the dry-snow zone
(Summit). The analysis of the NASA GISS ModelE AeroCom outputs shows no
statistically significant trend in the modelled fluxes for either location,
consistent with the results recently reported by Polashenski et al. (2015)
for the dry-snow zone. Results of the analysis of fluxes over different
areas point to similar conclusions. Similar results are obtained when
considering the months of January, February, and March, when aerosol
concentration is expected to be higher. The results here presented
complement other studies (e.g. Stone et al., 2014), indicating that, since
the 1980s, atmospheric concentrations of BC measured at surface stations in
the Arctic have decreased, with variations attributed to changes in both
anthropogenic and natural aerosol and aerosol precursor emissions.
Mean summer values of AOD (550 nm) measured at three AERONET
(http://aeronet.gsfc.nasa.gov) Greenland sites based in Thule (northwest
Greenland; 77∘28′00′′ N, 69∘13′50′′ W),
Ittoqqortoormiit (east-central Greenland; 70∘29′07′′ N,
21∘58′00′′ W), and Kangerlussuaq during the period 2007–2013
(with the starting year ranging between 2007 and 2009, depending on the
site) are reported in Table 2, together with their standard deviations. None
of the stations show statistically significant trends in AOD, consistent
with the results of the analysis of the modelled deposition fluxes.
Same as Fig. 9 but for Summit station.
A recent study (Dumont et al., 2014) concluded that dust deposition has been
increasing over much of the GrIS and that this is driving lowered albedo
across the ice sheet. That conclusion was based on trends of an “impurity
index”, which is the ratio of the logarithm of albedo in the 545–565 nm
MODIS band (where LAI affect albedo) to the logarithm of albedo in the
841–876 nm band (where they do not). In the MODIS product used in the study by Dumont et
al. (2014), albedo values rely on removal of the effects of aerosols
in the atmosphere. In the Dumont et al. (2014) study, this correction was
made using simulations of atmospheric aerosols by the Monitoring Atmospheric
Composition and Climate (MACC) model. Their resulting impurity index
shows positive trends, and these are attributed in part (up to 30 %) to
increases in atmospheric aerosol not accounted for by the model, and the
remainder to increases in snow LAI. The latter is consistent with our
findings herein: that GrIS darkening is in part attributable to an increase
in the impurity content of surface snow. However, Dumont et al. (2014)
assume that this increase in surface snow LAI is a result of enhanced
deposition from the atmosphere. They do not account for the possibility that
positive trends in impurity content may instead be a result of
warming-driven in-snow processes. Indeed, their own table shows variable AOD
at AERONET stations in Greenland, but no trend over the period studied (2007–2012).
The results of the analysis discussed above reinforce our argument that the
decline in the visible albedo over Greenland is probably not due to an
increase in the rate of deposition of LAI from the atmosphere, but instead
is due to the consolidation of LAI at the snow surface with warming-driven
increases in melt and/or sublimation and with the increased exposure of
underlying dirty ice.
Albedo projections through to 2100
We estimated future projections of summer albedo over the GrIS using MAR
forced with the outputs of three different Earth system models (ESMs) from
CMIP5, driven by two radiative forcing scenarios (Meinshausen et al., 2011)
over the 120-year period 1980–2100. The first scenario corresponds to an
increase in the atmospheric greenhouse gas concentration to a level of
850 ppm CO2 equivalent (RCP45); the second scenario increases CO2
equivalent to > 1370 ppm in 2100 (RCP85) (Moss et al., 2010;
Meinshausen et al., 2011). The three ESMs used are the second generation of
the Canadian Earth System Model (CanESM2), the Norwegian Community Earth
System Model (NorESM1), and the Model for Interdisciplinary Research on
Climate (MIROC5) of the University of Tokyo, Japan. More information is
available in Tedesco and Fettweis (2012). The ESMs are used to generate MAR
outputs for the historical period (1980–2005) and for future projections
(2005–2100). The Canadian Earth System Model (CanESM2, e.g. Arora and
Boer, 2010; Chylek et al., 2011) combines the fourth-generation climate
model (CanCM4) from the Canadian Centre for Climate Modelling and Analysis
with the terrestrial carbon cycle based on the Canadian Terrestrial
Ecosystem Model (CTEM), which models the land–atmosphere carbon exchange.
The NorESM1 model is built under the structure of the Community Earth System
Model (CESM) of the National Center for Atmospheric Research (NCAR). The
major difference from the standard CESM configuration concerns a
modification to the treatment of atmospheric chemistry, aerosols, and clouds
(Seland et al., 2008) and the ocean component. Lastly, MIROC5 is a coupled
general circulation model developed at the Center for Climate System
Research (CCSR) of the University of Tokyo, composed of the CCSR/NIES
(National Institute of Environmental Studies) atmospheric general
circulation model (AGCM 5.5) and the CCSR Ocean Component Model, including a
dynamic–thermodynamic sea-ice model (e.g. Watanabe et al., 2010, 2011). We
refer to Tedesco and Fettweis (2012) for the evaluation of the outputs of
MAR when forced with the outputs of the ESMs during the historical period
(1980–2005). All simulations consistently point to darkening accelerating
through the end of the century (Fig. 11), with summer albedo anomalies
(relative to the year 2000) as large as -0.08 by the end of the century over the
whole ice sheet, and even greater (-0.1) over the western portion of the ice
sheet (Fig. 12). The magnitude of the projected albedo anomalies by 2100,
however, is probably underestimated by our simulations, because (a) the
model tends to underestimate melting when forced with the ESMs (Fettweis et
al., 2013), and therefore underestimates grain size growth, and (b) the
model currently does not account for the presence of LAI in the snow or on
the ice surface, nor for the positive feedback between LAI and snow/ice melt.
Projections of broadband albedo anomaly (with respect to the year
2000) averaged over the whole GrIS for 1990–2012 from MAR simulations and
GLASS retrievals (black and red lines, respectively), and as projected by
2100. Future projections are simulated with MAR forced at its boundaries
with the outputs of three ESMs under two warming scenarios, with the first
scenario (RCP45) corresponding to an increase in the atmospheric greenhouse
gas concentration to a level of 850 ppm CO2 equivalent by 2100 and the
second (RCP85) to > 1370 pm CO2 equivalent. The top and the
bottom of the coloured area plots represent the results concerning the RCP45
(top panels) and RCP85 (bottom panels) scenarios. Semi-transparent colours are used to
allow overlapping data to be viewed. Dark green corresponds to the case where
MIROC5 and CANESM2 results overlap and brown to the case when the results
from the three ESMs overlap.
Same as Fig. 11 but for different drainage regions of the GrIS,
indicated by the small maps in each panel. The colour scheme for the shaded
regions is the same as Fig. 10. The top and the bottom of each area plots
represent the results concerning the RCP45 (bottom panels) and RCP85 (top panels)
scenarios. Red lines represent the GLASS albedo averaged over the
corresponding drainage region.
Discussion
Our results show a darkening of the GrIS 1996–2012, and indicate that this
darkening is associated with increased surface snow grain size, an expansion
in the area and persistence of bare ice, and by an increase in surface snow
light-absorbing impurity (LAI) concentrations. We find no evidence for
general increases in the deposition of LAI across the GrIS, so we associate
the higher surface snow impurity concentrations predominantly with the
appearance of underlying dirty ice and the consolidation of LAI in surface
snow resulting from snow melt. Interannual variability in the JJA GLASS
albedo is captured by the MAR albedo simulations, with the latter explaining
∼ 90 % of the space-borne-derived albedo interannual
variations for the period 1996–2012. The strong correlation between MAR
and GLASS albedo time series for this period suggests that MAR is capturing
the processes driving most of the albedo interannual variability (grain
size metamorphism and bare ice exposure) and that these processes have more
influence than those associated with the spatial and temporal variability of
surface impurity concentrations at seasonal timescales (currently not
included in the MAR albedo scheme). This is reinforced by the fact that the
range of snow grain size found across the GrIS produces larger changes in
albedo than does the range of LAI concentrations measured over the GrIS, at
least in the cold-snow and percolation zones of the ice sheet. As pointed
out by Tedesco et al. (2015), for pure snow, grain growth from new snow
(with r= 100 µm) to old melting snow (r= 1000 µm) can reduce
broadband albedo by ∼ 10 %. By comparison, adding 20 ng g-1
of BC, which has been found in the top layer of melting GrIS snow,
reduces albedo by only 1–2 %, consistent with the results reported by
Polashenski et al. (2015).
Modelled (MAR) and retrieved (GLASS) albedo are compared, with the latter
showing stronger declines in GrIS albedo, particularly over the ablation
zone. Based on our analysis, we suggest that the difference between MAR and
GLASS trends cannot be driven solely by the MODIS sensor degradation on the
Terra satellite (also used in the GLASS product), because the estimated
impact of sensor degradation on the albedo trend is much smaller than the
difference between the MAR and GLASS trends, and because the GLASS product
is obtained by combining data from both Terra and Aqua satellites, hence
likely reducing the impact of the Terra sensor degradation on the trends.
This is especially true over the dark zone, where substantial melting
occurs and where the albedo decline is pronounced. As mentioned, a lack of
impurities in the MAR albedo scheme can affect simulated albedo trends in at
least two ways: first, the concentration of impurities over bare ice areas
could be increasing or/and the lack of impurities in the MAR albedo scheme
causes bare ice areas to have an overestimated albedo. Moreover, more
frequent exposure of bare ice would lead to a decline in annual average
albedo over time, with such a trend being larger in the case of the presence
of impurity concentrations on the ice surface. Our sensitivity analysis of
the simulated trends on the bare ice albedo value indicates that the
difference between MAR and GLASS estimated trends is consistent with a
relatively darker (e.g. containing LAI) bare ice. Since MAR does not
account for the presence of surface LAI, and because the impact of LAI is
mostly in the UV and visible portion of the spectrum, we suggest that
another mechanism explaining the difference of -0.017 decade-1 between
the MAR and GLASS visible albedo trends is associated with increasing
mixing ratios of LAI in surface snow and ice on some parts of the GrIS. As
we pointed out, this could be due to a combination of increased exposure of
dirty ice with ablation (Wientjes and Oerlemans, 2010; Bøggild et al.,
2010), to enhanced melt consolidation with warming (e.g. Doherty et al.,
2013), or to increased deposition of LAI from the atmosphere. The absence of
in situ spatially distributed measurements to separate these processes
means that we cannot quantify their relative contributions to the darkening
in the visible region. Based on our analysis of trends in AOD over Greenland
and the lack of a trend in forest-fire counts and BC in North America and
Eurasia, we argue that increased deposition of LAI is not a large driver for
the observed negative trends in Greenland surface albedo. An exception could
be an increase in the deposition of locally transported dust near the
glacial margins, which would primarily affect the ablation zone. In
particular, locally lofted dust may be playing a substantial role in the
southwest GrIS ablation zone. However, we note that increased deposition is
not needed in order to have an increase in the concentration of LAI at the
GrIS surface. As noted above, indeed, temperatures and melt rates have been
accelerating over the GrIS during the past decades (e.g. Tedesco et al.,
2014). When snow melts, snow water is removed from the surface more
efficiently than particulate impurities; the result is an increase in
impurity concentrations in surface snow (e.g. Flanner et al., 2007; Doherty
et al., 2013). Large particles, such as dust, in particular, will have poor
mobility through the snowpack (Conway et al., 1996) so their concentration
at the surface is expected to increase with snowmelt. This effect may be
especially amplifying snow impurity content in the low-altitude ablation
zone of the GrIS, where enhanced melting has been occurring (e.g. Tedesco
et al., 2014). Further, the albedo reduction for a given concentration of an
absorbing impurity in snow is greater in large-grained snow than in
small-grained snow (Fig. 7 of Warren and Wiscombe, 1980; Flanner et al.,
2007), so climate warming itself will amplify the effect of LAI on surface
albedo. Warming may also lead to increased sublimation, removing snow water
but not particles from the snow surface, again increasing concentrations of
LAI in surface snow.
May–June-averaged aerosol optical depth at 550 nm for (a) dust,
(b) organic matter, (c) black carbon, and (d) total obtained from the GOCART
model and from the MACC model (as in Dumont et al., 2014) for the domain
bounded by 75 to 80∘ N and 30 to 50∘ W. All trends are
not statistically significant, with the exception of the MACC outputs for
dust (p< 0.01) and total aerosol (p< 0.05).
Snow and ice warmed by increased temperatures and higher LAI concentrations
also promotes darkening via so-called “bio albedo”, with biological growth
on the surface depressing the albedo. Green, pink, purple, brown, and black
pigmented algae, indeed, occur in melting snow and ice. Microbes can bind to
particulates, including BC, retaining them at the surface in higher
concentrations than in the parent snow and ice. The magnitude of this source
of darkening is currently unquantified, but as the climate warms and melt
seasons lengthen, biological habitats are expected to expand, with their
contribution to darkening likely increasing (Benning et al., 2014).
Quantifying the impact of aerosols on Greenland darkening is also made
difficult by the large disagreements among models in their predicted aerosol
deposition rates over the GrIS. We examine the contrast between AOD trends
from the MACC model used by Dumont et al. (2014) and the Goddard Chemistry
Aerosol Radiation and Transport model (GOCART). The GOCART model simulates
major tropospheric aerosol components, including sulfate, dust, BC, organic
carbon, and sea-salt aerosols using assimilated meteorological fields
of the Goddard Earth Observing System Data Assimilation System (GEOS DAS),
generated by the Goddard Global Modeling and Assimilation Office. Figure 13
compares results for AOD at 550 nm from MACC and GOCART for dust, organic
matter, and black carbon for the domain bounded by 75 to 80∘ N and
30 to 50∘ W (the same area considered by Dumont et al., 2014). The
MACC model shows statistically significant trends for dust (p< 0.01)
and for total aerosols (p< 0.05). All remaining trends are not
statistically significant for both MACC and GOCART outputs (Fig. 13).
Neither model represents the process of increased exposed silt/dust as
Greenland glaciers recede; therefore, we would not expect them to capture
trends in dust from this source. The inconsistency between the MACC and
GOCART values and trends is puzzling, and indicates that the simulation of
aerosol deposition rates over Greenland needs improvement.
Summary and conclusions
We studied the mean summer broadband albedo over the Greenland ice sheet
between 1981 and 2012 as estimated from space-borne measurements and found
that summer albedo decreased at a rate of 0.02 decade-1 between 1996
and 2012. The analysis of the outputs of the MAR regional climate model
indicates that the observed darkening is associated with increasing
temperatures and enhanced melting occurring during the same period, which in
turn promote increased surface snow grain size as well as the expansion and
persistency of areas with exposed bare ice. The MAR model simulates the
interannual variability in the retrieved GLASS albedo well, but the albedo trend
is larger in the GLASS albedo product than in MAR, indicating that processes
not represented in the MAR physics account for some of the declining albedo.
Specifically, we suggest that the absence of the effects of light-absorbing
impurities in MAR could account for the difference. We also suggest that
this hypothesis is supported by the trends observed along the ablation zone,
where the differences between observed and modelled trends are more pronounced
and the effect of the Terra sensor degradation plays a relatively small
role. On the other hand, over the dry-snow zone, our hypothesis requires
further testing, in view of the potentially higher impact of the sensor
degradation on the observed albedo trend. The analysis of modelled fields
and in situ data indicated an absence of trends in aerosol optical depth
over Greenland, as well as no significant trend in particulate
light-absorbing emissions (e.g. BC) from fires in likely source regions.
This is consistent with the absence of trends in surface aerosol
concentrations measured around the Arctic. Consequently, we suggest that the
increased surface concentrations of LAI associated with the darkening are not
related to increased deposition of LAI, but rather to post-depositional
processes, including increased loss of snow water to sublimation and melt
and the outcropping of “dirty” underlying ice associated with snow/firn
removal due to ablation.
Future projections of GrIS albedo obtained from MAR forced under different
warming scenarios point to continued darkening through the end of the
century, with regions along the edges of the ice sheet subject to the
largest decrease, driven solely by warming-driven changes in snow grain
size, exposure of bare ice, and melt pool formation. We hypothesize that
projected darkening trends would be even greater in view of the
underestimated projected melting (and effect on albedo) and in view of the
fact that the current version of the MAR model does not account for the
presence of surface LAI and the associated positive direct and indirect
impact on lowered albedo.
The drivers we identified to be responsible for the observed darkening are
related to endogenous processes rather than exogenous ones and are strongly
driven by melting. Because melting is projected to increase over the next
decades, it is crucial to assess our capability of studying,
quantifying, and projecting these processes as they will inevitably impact,
and be impacted by, future scenarios. Intrinsic limitations of current
observational tools and techniques, the scarcity of in situ observations, and
the albedo schemes currently used in existing models of surface energy
balance and mass balance limit our ability to separate the contributions to
darkening by the different processes, especially with regard to the cause
and evolution of surface impurity concentrations. Moreover, as with all
instruments, sensors undergo deterioration, and it can be difficult to
separate an albedo trend from sensor drift. This is especially true in the
dry-snow zone, where impurity concentrations are extremely low (only a few
parts per billion (ppb) in the case of BC). In this regard, a recent study by Polashenski et al. (2015) suggests that the decline and spectral shift in dry-snow albedo over
Greenland contains important contributions from uncorrected Terra sensor
degradation when using the MODIS data collection C5. The new MODIS Terra
version (accounting for the sensor degradation) does not appear to show any
trend (C. Polashenski, personal communication, 2015), hence supporting the hypothesis of the absence of trends
of LAI deposition over the dry zone.
Remote sensing and in situ observations should be complemented with models
that simulate the surface energy balance to account for the evolution of the
snowpack, in particular changes in surface grain size and exposure of bare
ice. Simulations with regional climate models can provide such quantities,
but they do not currently account for the transport and deposition of LAI to
Greenland, the post-depositional evolution of impurities in the snowpack,
and the synergism between surface LAI and grain growth (whereby a given
impurity content causes more albedo reduction in coarse-grained snow than in
fine-grained snow). In this regard, the current parameterization for snow
albedo in MAR is based on that of Brun et al. (1992), as part of an
avalanche-forecasting model. As a consequence of the results of this study,
we began evaluating an alternative albedo scheme using a parameterization
that can also account for the albedo reduction by absorptive impurities
(e.g. Dang et al., 2015). Moreover, we are also considering using the
firn/ice albedo parameterization of Dadic et al. (2013), based on
measurements covering the range of densities from 400 to 900 kg m-3.
Surface-based measurements are needed to test satellite-retrieved albedo and
to quantify the drivers behind albedo changes in different areas of
Greenland. To date, most surface-based observations have been made in the
dry-snow zone or the percolation zone, and they have generally focused on
measuring the mixing ratios of BC (Hagler et al., 2007; McConnell et al.,
2007; Polashenski et al., 2015) or of the spectral light absorption by
all particulate components collectively (Doherty et al., 2010; Hegg et al.,
2009, 2010). The regions of Greenland that are darkening the most rapidly
are within the ablation zone. Here, there is no direct evidence that the
rate of atmospheric deposition of LAI has been increasing. In view of the
cumulative effect of snowmelt leaving impurities at the surface, the
intra-seasonal variation of deposition may not be as important as the
exposure of LAI by melting. Changes in the abundances of light-absorbing
algae and other organic material with warmer temperatures may also be
contributing to declining albedo, particularly for the ice, but this is an
essentially unstudied source of darkening. Until measurements are made that
quantify and distinguish the relative roles of each of these factors in the
darkening of the GrIS, it is not possible to reduce the uncertainty in their
contributions to the acceleration of surface melt. In addition to the need
for targeted ground observations, it is necessary for the models that
simulate and project the evolution of surface conditions over Greenland to
start including the contribution of surface LAI, their processes, and their
impact on albedo, as well as aerosol models that account for their
deposition. Concurrently, space-borne sensors or missions capable of
separating the contributions from the different processes (with increased
spatial, spectral, and radiometric resolution) should be planned for remote
sensing to become a more valuable tool in this regard.
M. Tedesco conceived the study, carried out the scientific analysis, and wrote the
main body of the manuscript. S. Doherty co-wrote the manuscript and provided
feedback on the analysis of the impact of surface LAI on the albedo
decrease. P. Alexander provided MODIS visible data for the comparison with the
GLASS-estimated visible albedo. J. Jeyaratnam supported the reprojection and analysis
of GLASS and MAR data. X. Fettweis contributed with the analysis of MAR outputs. M. Tedesco,
S. Doherty, X. Fettweis, and J. Stroeve edited the final version of the manuscript.
Acknowledgements
M. Tedesco and P. Alexander were supported by NSF grants PLR1304807 and ANS 0909388, and NASA
grant NNX1498G. The authors are grateful to Kostas Tsirigadis (NASA GISS)
for providing the outputs of GISS modelE of the AeroCom phase II project and
to Marie Dumont, Eric Brun, and Samuel Morin for the data used in Fig. 13.
We thank Tao He at the University of Maryland, College Park, for the
discussion on the GLASS product. The authors thank Stephen Warren for
providing suggestions and guidance during the preparation of the manuscript,
particularly for pointing out limitations and providing suggestions on the
albedo parameterizations.
Edited by: E. Hanna
ReferencesAlexander, P. M., Tedesco, M., Fettweis, X., van de Wal, R. S. W., Smeets,
C. J. P. P., and van den Broeke, M. R.: Assessing spatio-temporal variability
and trends in modelled and measured Greenland Ice Sheet albedo (2000–2013),
The Cryosphere, 8, 2293–2312, 10.5194/tc-8-2293-2014, 2014.
Arnaud, L., Barnola, J. M., and Duval, P.: Physical modeling of the
densification of snow/firn and ice in the upper part of polar ice sheets, in:
Physics of Ice Core Records, edited by: Hondoh, T., Hokkaido University Press,
Sapporo, Japan, 285–305, 2000.Arora, V. K. and Boer, G. J.: Uncertainties in the 20th century carbon budget
associated with land use change, Global Change Biol., 16, 3327–3348,
10.1111/j.1365-2486.2010.02202.x, 2010.Benning, L. G., Anesio, A. M., Lutz, S., and Tranter, M.: Biological impact
on Greenland's albedo, Nat. Geosci., 7, 691–691, 10.1038/ngeo2260, 2014.
Benson, C. S.: Stratigraphic Studies in the Snow and Firn of the Greenland Ice
Sheet, Research Report 70, US Army Snow, Ice, and Permafrost Research
Establishment (SIPRE), US Army Cold Regions Research and Engineering Laboratory (CRREL),
published version of C. S. Benson's PhD Dissertation of 1960, 93 pp., 1962.
Bøggild, C. E., Brandt, R. E., Brown, K. J., and Warren, S. G.: The
ablation zone in northeast Greenland: Ice types, albedos, and impurities,
J. Glaciol., 56, 101–113, 2010.Bond, T. C., Doherty, S. J., Fahey, D. W., Forster, P. M., Berntsen, T.,
DeAngelo, B. J., Flanner, M. G., Ghan, S., Kärcher, B., Koch, D., Kinne,
S., Kondo, Y., Quinn, P. K., Sarofim, M. C., Schultz, M. G., Schulz, M.,
Venkataraman, C., Zhang, H., Zhang, S., Bellouin, N., Guttikunda, S. K.,
Hopke, P. K., Jacobon, M. Z., Kaiser, J. W., Klimont, Z., Lohmann, U., Schwarz,
J. P., Shindell, D., Storelvmo, T., Warren, S. G., and Zender, C. S.: Bounding the Role
of Black Carbon in Climate: A scientific assessment, J. Geophys. Res.,
118, 5380–5552, 10.1002/jgrd.50171, 2013.Box, J. E., Fettweis, X., Stroeve, J. C., Tedesco, M., Hall, D. K., and
Steffen, K.: Greenland ice sheet albedo feedback: thermodynamics and
atmospheric drivers, The Cryosphere, 6, 821–839, 10.5194/tc-6-821-2012, 2012.
Brun, E., David, P., Sudul, M., and Brunot, G.: A numerical model to simulate
snow-cover stratigraphy for operational avalanche forecasting, J. Glaciol., 38, 13–22, 1992.
Cachier, H. and Pertuisot, M. H.: Particulate carbon in Arctic ice, Analusis
Mag., 22, 34–37, 1994.
Chýlek, P., Johnson, B., Damiano, P. A., Taylor, K. C., and
Clement, P.: Biomass burning record and black carbon in the GISP2 ice core,
Geophys. Res. Lett., 22, 89–92, 1995.Chylek, P., Li, J., Dubey, M. K., Wang, M., and Lesins, G.: Observed and model
simulated 20th century Arctic temperature variability: Canadian Earth System
Model CanESM2, Atmos. Chem. Phys. Discuss., 11, 22893–22907, 10.5194/acpd-11-22893-2011, 2011.
Clarke, A. D. and Noone, K. J.: Soot in the Arctic snowpack: A cause for
perturbations in radiative transfer, Atmos. Environ., 19, 2045–2053, 1985.
Conway, H., Gades, A., and Raymond, C. F.: Albedo of dirty snow during
conditions of melt, Water Resour. Res., 32, 1713–1718, 1996.Dadic, R., Mullen, P. C., Schneebeli, M., Brandt, R. E., and Warren, S. G.:
Effects of bubbles, cracks, and volcanic tephra on the spectral albedo of
bare ice near the Trans-Antarctic Mountains: implications for sea-glaciers
on Snowball Earth, J. Geophys. Res.-Earth, 118, 1658–1676, 10.1002/jgrf.20098, 2013.Dang, C., Brandt, R. E., and Warren, S. G.: Parameterizations for
narrowband and broadband albedo of pure snow, and snow containing mineral
dust and black carbon, J. Geophys. Res., 120, 5446–5468, 10.1002/2014JD022646, 2015.
De Ridder, K. and Galleé, H.: Land surface-induced regional
climate change in Southern Israel, Appl. Meteorol., 37, 1470–1485, 1998.Doherty, S. J., Warren, S. G., Grenfell, T. C., Clarke, A. D., and Brandt, R. E.:
Light-absorbing impurities in Arctic snow, Atmos. Chem. Phys., 10, 11647–11680,
10.5194/acp-10-11647-2010, 2010.
Doherty, S. J., Grenfell, T. C., Forsström, S., Hegg, D. L., Brandt, R.
E., and Warren, S. G.: Observed vertical redistribution of black carbon and
other insoluble light-absorbing particles in melting snow, J. Geophys. Res.-Atmos.,
118, 5553–5569, 2013.
Dumont, M., Brun, E., Picard, G., Michou, M., Libois, Q., Petit, J.-R.,
Geyer, M., Morin, S., and Josse, B.: Contribution of light-absorbing
impurities in snow to Greenland's darkening since 2009, Nat. Geosci., 7, 509–512, 2014.
Fettweis, X., Gallée, H., Lefebre, F., and van Ypersele, J.-P.: Greenland
surface mass balance simulated by a regional climate model and comparison
with satellite-derived data in 1990–1991, Clim. Dynam., 24, 623–640, 2005.Fettweis, X., Franco, B., Tedesco, M., van Angelen, J. H., Lenaerts, J. T. M.,
van den Broeke, M. R., and Gallée, H.: Estimating the Greenland ice sheet surface
mass balance contribution to future sea level rise using the regional atmospheric
climate model MAR, The Cryosphere, 7, 469–489, 10.5194/tc-7-469-2013, 2013.Flanner, M. G., Zender, C. S., Randerson, J. T., and Rasch, P. J.:
Present-day climate forcing and response from black carbon in snow, J.
Geophys. Res., 112, D11202, 10.1029/2006JD008003, 2007.
Grenfell, T. C., Perovich, D. K., and Ogren, J. A.: Spectral albedos of an
alpine snowpack, Cold Reg. Sci. Technol., 4, 121–127, 1981.
Greuell, W. and Konzelman, T.: Numerical modelling of the energy balance and
the englacial temperature of the Greenland Ice Sheet, Calculations for the
ETH-Camp location (West Greenland, 1155 m a.s.l.), Global Planet. Change,
9, 91–114, 1994.Hagler, G. S. W., Bergin, M. H., Smith, E. A., and Dibb, J. E.: A summer
time series of particulate carbon in the air and snow at Summit, Greenland,
J. Geophys. Res., 112, D21309, 10.1029/2007JD008993, 2007.He, T., Liang, S., Yu, Y., Wang, D., Gao, F., and Liu, Q.: Greenland surface
albedo changes in July 1981–2012 from satellite observations, Environ. Res.
Lett., 8, 044043, 10.1088/1748-9326/8/4/044043, 2013.
Hegg, D. A., Warren, S. G., Grenfell, T. C., Doherty, S. J., Larson, T. V.,
and Clarke, A. D.: Source Attribution of Black Carbon in Arctic Snow,
Environ. Sci. Technol., 43, 4016–4021, 2009.Hegg, Dean A., Warren, Stephen G., Grenfell, Thomas C., Sarah J Doherty, and
Clarke, Antony D.: Sources of light-absorbing aerosol in arctic snow and their
seasonal variation, Atmos. Chem. Phys., 10, 10923–10938, 10.5194/acp-10-10923-2010, 2010.
Herron, M. M. and Langway Jr., C. C.: Firn densification: An empirical model,
J. Glaciol., 25, 373–385, 1980.Jiao, C., Flanner, M. G., Balkanski, Y., Bauer, S. E., Bellouin, N., Berntsen,
T. K., Bian, H., Carslaw, K. S., Chin, M., De Luca, N., Diehl, T., Ghan, S. J.,
Iversen, T., Kirkevåg, A., Koch, D., Liu, X., Mann, G. W., Penner, J. E.,
Pitari, G., Schulz, M., Seland, Ø., Skeie, R. B., Steenrod, S. D., Stier, P.,
Takemura, T., Tsigaridis, K., van Noije, T., Yun, Y., and Zhang, K.: An AeroCom
assessment of black carbon in Arctic snow and sea ice, Atmos. Chem. Phys., 14,
2399–2417, 10.5194/acp-14-2399-2014, 2014.Keegan, K. M., Albert, M. R., McConnell, J. R., and Baker, I.: Climate change
and forest fires synergistically drive widespread met events of the
Greenland Ice Sheet, P. Natl. Acad. Sci. USA, 111, 7964–7967, 10.1073/pnas.1405397111, 2014.
LaChapelle, E. R.: Field Guide to Snow Crystals, University of Washington
Press, Seattle, 1969.Lefebre, F., Gallée, H., and van Ypersele, J.-P.: Modeling of snow and
ice melt at ETH Camp (West Greenland): a study of surface albedo, J. Geophys.
Res.-Atmos., 108, 4231, 10.1029/2001JD001160, 2003.
Liang, S., Zhao, X., Liu, S., Yuan, W., Cheng, X., Xiao, Z., Zhang, X., Liu, Q.,
Cheng, J., Tang, H., Qu, Y., Bo, Y., Qu, Y., Ren, H., Yu, K., and Townshend, J.:
A long-term Global LAnd Surface Satellite (GLASS) data-set for environmental
studies, Int. J. Digit. Earth, 6, 5–33, 2013.McConnell, J. R., Edwards, R., Kok, G. L., Flanner, M. G., Zender, C. S.,
Saltzman, E. S., Banta, J. R., Pasteris, D. R., Carter, M. M., and Kahl,
J. D. W.: 20th-century industrial black carbon emissions altered Arctic climate
forcing, Science, 317, 1381–1384, 10.1126/science.1144856, 2007.
Meinshausen, M., Smith, S. J., Calvin, K., Daniel, J. S., Kainuma, M. L. T.,
Lamarque, J.-F. Matsumoto, K., Montzka, S. A., Raper, S. C. B., Riahi, K.,
Thomson, A., Velders, G. J. M., and Vuuren, D. P. P.: The RCP greenhouse gas concentrations and their
extensions from 1765 to 2300, Climatic Change, 109, 213–241, 2011.
Moss, R. H., Edmonds, J. A., Hibbard, K. A., Manning, M. R., Rose, S. K., van
Vuuren, D. P., Carter, T. R., Emori, S., Kainuma, M., Kram, T., Meehl, G. A.,
Mitchell, J. F. B., Nakicenovic, N. Riahi, K., Smith, S. J., Stouffer, R. J.,
Thomson, A. M., Weyant, J. P., and Wilbanks, T. J.: The next generation of scenarios for climate change
research and assessment, Nature, 463, 747–756, 2010.Myhre, G., Samset, B. H., Schulz, M., Balkanski, Y., Bauer, S., Berntsen, T. K.,
Bian, H., Bellouin, N., Chin, M., Diehl, T., Easter, R. C., Feichter, J., Ghan,
S. J., Hauglustaine, D., Iversen, T., Kinne, S., Kirkevåg, A., Lamarque,
J.-F., Lin, G., Liu, X., Lund, M. T., Luo, G., Ma, X., van Noije, T., Penner,
J. E., Rasch, P. J., Ruiz, A., Seland, Ø., Skeie, R. B., Stier, P., Takemura,
T., Tsigaridis, K., Wang, P., Wang, Z., Xu, L., Yu, H., Yu, F., Yoon, J.-H.,
Zhang, K., Zhang, H., and Zhou, C.: Radiative forcing of the direct aerosol
effect from AeroCom Phase II simulations, Atmos. Chem. Phys., 13, 1853–1877,
10.5194/acp-13-1853-2013, 2013.Nghiem, S. V., Hall, D. K., Mote, T. L., Tedesco, M., Albert, M. R., Keegan,
K., Shuman, C. A., DiGirolamo, N. E., and Neumann, G.: The extreme melt
across the Greenland ice sheet in 2012, Geophys. Res. Lett., 39, L20502,
10.1029/2012GL053611, 2012.Polashenski, C. M., Dibb, J. E., Flanner, M. G., Chen, J. Y., Courville, Z.
R., Lai, A. M., Schauer, J. J., Shafer, M. M., and Bergin, M.: Neither dust nor
black carbon causing apparent albedo decline in Greenland's dry snow zone:
Implications for MODIS C5 surface reflectance, Geophys. Res. Lett., 42,
9319–9327, 10.1002/2015GL065912, 2015.Rae, J. G. L., Aðalgeirsdóttir, G., Edwards, T. L., Fettweis, X.,
Gregory, J. M., Hewitt, H. T., Lowe, J. A., Lucas-Picher, P., Mottram, R. H.,
Payne, A. J., Ridley, J. K., Shannon, S. R., van de Berg, W. J., van de Wal,
R. S. W., and van den Broeke, M. R.: Greenland ice sheet surface mass balance:
evaluating simulations and making projections with regional climate models,
The Cryosphere, 6, 1275–1294, 10.5194/tc-6-1275-2012, 2012.Samset, B. H., Myhre, G., Herber, A., Kondo, Y., Li, S.-M., Moteki, N., Koike,
M., Oshima, N., Schwarz, J. P., Balkanski, Y., Bauer, S. E., Bellouin, N.,
Berntsen, T. K., Bian, H., Chin, M., Diehl, T., Easter, R. C., Ghan, S. J.,
Iversen, T., Kirkevåg, A., Lamarque, J.-F., Lin, G., Liu, X., Penner, J. E.,
Schulz, M., Seland, Ø., Skeie, R. B., Stier, P., Takemura, T., Tsigaridis,
K., and Zhang, K.: Modelled black carbon radiative forcing and atmospheric
lifetime in AeroCom Phase II constrained by aircraft observations, Atmos.
Chem. Phys., 14, 12465–12477, 10.5194/acp-14-12465-2014, 2014.
Schaaf, C. B., Gao, F., Strahler, A. H., Lucht, W., Li, X., Tsang, T.,
Strugnell, N. C., Zhang, X., Jin, Y., Muller, J. P., Lewis, P., Barnsley,
M., Hobson, P., Disney, M., Roberts, G., Dunderdale, M., Doll, C.,
d'Entremont, R. P., Hu, B., Liang, S., Privette, J. L., and Roy, D.: First
operational BRDF, albedo nadir reflectance products from MODIS, Remote
Sens. Environ., 115, 1296–1300, 2002.Seland, O., Iversen, T., Kirkevag, A., and Storelvmo, T.: Aerosol-climate
interactions in the CAM-Oslo atmospheric GCM and investigation of associated
basic shortcomings, Tellus A, 60, 459–491, 10.1111/j.1600-0870.2008.00318.x, 2008.
Shepherd, A., Ivins, E. R. A. G., Barletta, V. R., Bentley, M. J., Bettadpur, S.,
Briggs, K. H., Bromwich, D. H., Forsberg, R., Galin, N., Horwath, M., Jacobs, S.,
Joughin, I., King, M. A., Lenaerts, J. T. M., Li, J., Ligtenberg, S. R. M.,
Luckman, A., Luthcke, S. B., McMillan, M., Meister, R., Milne, G., Mouginot, J.,
Muir, A., Nicolas, J., Paden, J., Payne, A. J., Pritchard, H., Rignot, E., Rott,
H., Sørensen, L., Scambos, T. A., Scheuchl, B., Schrama, E. J. O., Smith, B.,
Sundal, A. V., van Angelen, J. H., van de Berg, W. J., van den Broeke, M. R.,
Vaughan, D. G., Velicogna, I., Wahr, J., Whitehouse, P. L., Wingham, D. J.,
Yi, D., Young, D., and Zwally, H. J.: A Reconciled Estimate of Ice-Sheet Mass Balance,
Science, 338, 1183–1189, 2012.
Steffen, K. and Box, J. E.: Surface climatology of the Greenland ice sheet:
Greenland Climate Network 1995–1999, J. Geophys. Res., 106, 33951–33964, 2001.Stone, R. S., Sharma, S., Herber, A., Eleftheriadis, K., and Nelson, D. W.: A
characterization of Arctic aerosols on the basis of aerosol optical depth
and black carbon measurements, Elementa, 2, 000027, 10.12952/journal.elementa.000027, 2014.
Stroeve, J., Box, J. E., Gao, F., Liang, S., Nolin, A., and Schaaf, C.:
Accuracy assessment of the MODIS 16-day albedo product for snow: comparisons
with Greenland in situ measurements, Remote Sens. Environ., 94, 46–60, 2005.
Stroeve, J., Box, J. E., Wang, Z., Schaaf, C., and Barrett, A.: Re-evaluation
of MODIS MCD43 Greenland albedo accuracy and trends, Remote Sens. Environ.,
138, 199–214, 2013.Tedesco, M. and Fettweis, X.: 21st century projections of surface mass
balance changes for major drainage systems of the Greenland ice sheet,
Environ. Res. Lett., 7, 045405, 10.1088/1748-9326/7/4/045405, 2012.Tedesco, M., Fettweis, X., Broeke, M. R. V. D., de Wal, R. S. W. V., Smeets,
C. J. P. P., de Berg, W. J. V., Serreze, M. C., and Box, J. E.: The role of
albedo and accumulation in the 2010 melting record in Greenland, Environ.
Res. Lett., 6, 014005, 10.1088/1748-9326/6/1/014005, 2011.Tedesco, M., Fettweis, X., Mote, T., Wahr, J., Alexander, P., Box, J. E., and
Wouters, B.: Evidence and analysis of 2012 Greenland records from spaceborne
observations, a regional climate model and reanalysis data, The Cryosphere,
7, 615–630, 10.5194/tc-7-615-2013, 2013.
Tedesco, M., Box, J. E., Cappelen, J., Fettweis, X., Mote, T., Rennermalm, A.
K., van de Wal, R. S. W., and Wahr, J.: Greenland Ice Sheet in [2013 NOAA Arctic Report
Card], NOAA, 2014.Tedesco, M., Doherty, S., Warren, W., Tranter, M., Stroeve, J., Fettweis, X.,
and Alexander, P.: What darkens the Greenland Ice Sheet?, Eos, 96, 10.1029/2015EO035773, 2015.Tsigaridis, K., Daskalakis, N., Kanakidou, M., Adams, P. J., Artaxo, P., Bahadur,
R., Balkanski, Y., Bauer, S. E., Bellouin, N., Benedetti, A., Bergman, T.,
Berntsen, T. K., Beukes, J. P., Bian, H., Carslaw, K. S., Chin, M., Curci, G.,
Diehl, T., Easter, R. C., Ghan, S. J., Gong, S. L., Hodzic, A., Hoyle, C. R.,
Iversen, T., Jathar, S., Jimenez, J. L., Kaiser, J. W., Kirkevåg, A., Koch,
D., Kokkola, H., Lee, Y. H, Lin, G., Liu, X., Luo, G., Ma, X., Mann, G. W.,
Mihalopoulos, N., Morcrette, J.-J., Müller, J.-F., Myhre, G., Myriokefalitakis,
S., Ng, N. L., O'Donnell, D., Penner, J. E., Pozzoli, L., Pringle, K. J.,
Russell, L. M., Schulz, M., Sciare, J., Seland, Ø., Shindell, D. T., Sillman,
S., Skeie, R. B., Spracklen, D., Stavrakou, T., Steenrod, S. D., Takemura, T.,
Tiitta, P., Tilmes, S., Tost, H., van Noije, T., van Zyl, P. G., von Salzen,
K., Yu, F., Wang, Z., Wang, Z., Zaveri, R. A., Zhang, H., Zhang, K., Zhang, Q.,
and Zhang, X.: The AeroCom evaluation and intercomparison of organic aerosol
in global models, Atmos. Chem. Phys., 14, 10845–10895, 10.5194/acp-14-10845-2014, 2014.van Angelen, J. H., Lenaerts, J. T. M., Lhermitte, S., Fettweis, X., Kuipers
Munneke, P., van den Broeke, M. R., van Meijgaard, E., and Smeets, C. J. P. P.:
Sensitivity of Greenland Ice Sheet surface mass balance to surface albedo
parameterization: a study with a regional climate model, The Cryosphere, 6,
1175–1186, 10.5194/tc-6-1175-2012, 2012.van den Broeke, M. R., Smeets, C. J. P. P., and van de Wal, R. S. W.: The
seasonal cycle and interannual variability of surface energy balance and
melt in the ablation zone of the west Greenland ice sheet, The Cryosphere, 5,
377–390, 10.5194/tc-5-377-2011, 2011.Vernon, C. L., Bamber, J. L., Box, J. E., van den Broeke, M. R., Fettweis, X.,
Hanna, E., and Huybrechts, P.: Surface mass balance model intercomparison for
the Greenland ice sheet, The Cryosphere, 7, 599–614, 10.5194/tc-7-599-2013, 2013.
Wang, D., Morton, D., Masek, J., Wu, A., Nagol, J., Xiong, X., Levy, R.,
Vermote, E., and Wolfe, R.: Impact of sensor degradation on the MODIS NDVI
time series, Remote Sens. Environ., 119, 55–61, 2012.
Warren, S. G.: Optical properties of snow, Rev. Geophys. Space Phys., 20, 67–89, 1982.
Warren, S. G.: Can black carbon in snow be detected by remote sensing?,
J. Geophys. Res.-Atmos., 118, 779–786, 2013.
Warren, S. G., and Wiscombe, W. J.: A model for the spectral albedo of
snow, II: Snow containing atmospheric aerosols, J. Atmos. Sci., 37, 2734–2745, 1980.Watanabe, M., Suzuki, T., O'ishi, R., Komuro, Y., Watanabe, S., Emori, S.,
Takemura, T., Chikira, M., Ogura, T., Sekiguchi, M., Takata, K., Yamazaki, D.,
Yokohata, T., Nozawa, T., Hasumi, H., Tatebe, H., and Kimoto, M.: Improved Climate Simulation by MIROC5: Mean States,
Variability, and Climate Sensitivity, J. Climate, 23, 6312–6335, 2010.
Watanabe, S., Hajima, T., Sudo, K., Nagashima, T., Takemura, T., Okajima, H.,
Nozawa, T., Kawase, H., Abe, M., Yokohata, T., Ise, T., Sato, H., Kato, E.,
Takata, K., Emori, S., and Kawamiya, M.: MIROC-ESM 2010: model description and
basic results of CMIP5-20c3m experiments, Geosci. Model Dev., 4, 845–872,
10.5194/gmd-4-845-2011, 2011.Wientjes, I. G. M. and Oerlemans, J.: An explanation for the dark region in the
western melt zone of the Greenland ice sheet, The Cryosphere, 4, 261–268,
10.5194/tc-4-261-2010, 2010.Wientjes, I. G. M., Van de Wal, R. S. W., Reichart, G. J., Sluijs, A., and
Oerlemans, J.: Dust from the dark region in the western ablation zone of the
Greenland ice sheet, The Cryosphere, 5, 589–601, 10.5194/tc-5-589-2011, 2011.Xing, J., Pleim, J., Mathur, R., Pouliot, G., Hogrefe, C., Gan, C.-M., and
Wei, C.: Historical gaseous and primary aerosol emissions in the United States
from 1990 to 2010, Atmos. Chem. Phys., 13, 7531–7549, 10.5194/acp-13-7531-2013, 2013.Xing, J., Mathur, R., Pleim, J., Hogrefe, C., Gan, C.-M., Wong, D. C., Wei, C.,
Gilliam, R., and Pouliot, G.: Observations and modeling of air quality trends
over 1990–2010 across the Northern Hemisphere: China, the United States and
Europe, Atmos. Chem. Phys., 15, 2723–2747, 10.5194/acp-15-2723-2015, 2015.Ying, Q., Qiang, L., Shunlin, L., Lizhao, W., Nanfeng, L., and Suhong, L.:
Direct-Estimation Algorithm for Mapping Daily Land-Surface Broadband Albedo
From MODIS Data, IEEE T. Geosci. Remote, 52, 907–919, 10.1109/TGRS.2013.2245670, 2014.
Zhao, X., Liang, S., Liu, S., Yuan, W., Xiao, Z., Liu, Q., Cheng, J., Zhang,
X., Tang, H., Zhang, X., Liu, Q., Zhou, G., Xu, S., and Yu, K.: The Global Land
Surface Satellite (GLASS) Remote Sensing Data Processing System and
Products, Remote Sens., 5, 2436–2450, 2013.