SERIES A
DYNAMIC
METEOROLOGY
AND OCEANOGRAPHY
PUBLISHED BY THE INTERNATIONAL METEOROLOGICAL INSTITUTE IN STOCKHOLM
Arctic amplification: does it impact the polar jet stream?
By V A L EN T I N P . M E L ES H K O 1 * , O L A M . JO H A N N E S S E N 2 , 3 , A N D R E Y V . B A I D I N 1 ,
TAT IA NA V . P AVLOVA 1 a n d VE R O N I K A A . GO V O R K O V A 1 , 1Voeikov Main Geophysical
Observatory, St. Petersburg, Russia; 2Nansen Environmental and Remote Sensing Center, N-5006 Bergen,
Norway; 3Nansen Scientific Society, N-5006 Bergen, Norway
(Manuscript received 19 May 2016; in final form 24 August 2016)
ABSTRACT
It has been hypothesised that the Arctic amplification of temperature changes causes a decrease in the northward
temperature gradient in the troposphere, thereby enhancing the oscillation of planetary waves leading to extreme
weather in mid-latitudes. To test this hypothesis, we study the response of the atmosphere to Arctic amplification
for a projected summer sea-ice-free period using an atmospheric model with prescribed surface boundary
conditions from a state-of-the-art Earth system model. Besides a standard global warming simulation, we also
conducted a sensitivity experiment with sea ice and sea surface temperature anomalies in the Arctic. We show
that when global climate warms, enhancement of the northward heat transport provides the major contribution
to decrease the northward temperature gradient in the polar troposphere in cold seasons, causing more
oscillation of the planetary waves. However, while Arctic amplification significantly enhances near-surface air
temperature in the polar region, it is not large enough to invoke an increased oscillation of the planetary waves.
Keywords: arctic amplification, modelling, jet stream, surface air temperature
To access the supplementary material to this article, please see Supplementary files under
‘Article Tools’.
1. Introduction
One of the most outstanding manifestations of climate
change over the recent decades is the reduction of sea ice
in the Arctic. Satellite observations have shown a rapid
decline, particularly in September when the summer sea-ice
cover is minimum (Johannessen et al., 2004; Kattsov et al.,
2010; Stroeve et al., 2011; Comiso, 2012; Overland and
Wang, 2013; Vihma, 2014). September sea-ice extent and
area have decreased 8.8 and 10.4% per decade, respectively, since 1978 (Nansen Center Sea Ice Information
System, Bergen, Norway, www.arctic-roos.org). There is a
general consensus that under the current rate of CO2
increase, the Arctic Ocean will be ice-free during September
in the latter half of the 21st century (Johannessen et al.,
2004; Overland and Wang, 2010; Mahlstein and Knutti,
2012; Snape and Forster, 2014). Observational studies
indicate that Arctic surface temperature has increased twice
as fast as global average. This phenomenon commonly
known as Arctic amplification is mainly associated with
*Corresponding author.
email: [email protected]
global warming in recent decades, primarily caused by the
decline of sea ice (Screen et al., 2012). Recent studies have
also revealed that the Atlantic Multidecadal Oscillation
and Pacific Decadal Oscillation can modulate different
atmospheric responses to sea-ice decline (Johannessen et al.,
2016; Screen and Francis, 2016). One of the key issues
in the current debate is to understand the impact of Arctic
amplification on atmospheric circulation and occurrence
of extreme weather over Eurasia, Greenland and North
America (Budikova, 2009; Cohen et al., 2014; Vihma, 2014;
Overland et al., 2015). Based on a 30-yr observational
record, Francis and Vavrus (2012, 2015) hypothesised that
Arctic amplification is causing a decrease of northward
temperature gradient between the lower latitude and the
polar atmosphere, which favours weakening of upper-level,
flow, and increased meandering of planetary-scale waves.
Then the phase speed of such waves reduces due to their
larger amplitudes, resulting in a slowdown of wave propagation and increasing probabilities of extreme weather.
If the sea-ice cover continues to decrease as projected,
an important question is whether the large-scale circulation patterns in the Northern Hemisphere will become
Tellus A 2016. # 2016 V. P. Meleshko et al. This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 International
License (http://creativecommons.org/licenses/by/4.0/), allowing third parties to copy and redistribute the material in any medium or format and to remix,
transform, and build upon the material for any purpose, even commercially, provided the original work is properly cited and states its license.
Citation: Tellus A 2016, 68, 32330, http://dx.doi.org/10.3402/tellusa.v68.32330
1
(page number not for citation purpose)
2
V. P. MELESHKO ET AL.
increasingly influenced by Arctic amplification. Although
some aspects of the Francis and Vavrus (2012) hypothesis
are supported, the impact of Arctic amplification on the
decrease of northward temperature gradient and associated
growth of the planetary wave oscillation remains controversial and requires further investigation (Honda et al.,
2009: Screen et al., 2012; Barnes, 2013; Mori et al., 2014).
The quantitative cause-and-effect linkage between Arctic
change and mid-latitude weather cannot be resolved within
the foreseeable future based on the limited observational
record (Overland et al., 2015); therefore, numerical modelling
is a way forward.
The goal of our modelling study is to determine the
response of the polar and mid-latitude atmosphere to
anthropogenic warming when a sea-ice-free state is projected
to occur in the Arctic. The focus is on assessing the contribution of Arctic amplification and northward heat transport to changes of atmospheric variability to answer the
debated question: Is the Arctic amplification forcing large
enough to decrease the northward temperature gradient and
to enhance the oscillation of planetary-scale waves and to
impact regional atmospheric circulation including extreme
weather in the future? In our study, we compare a projected
sea-ice-free period in summer with the present condition in
the Arctic, because interaction processes between the Arctic
Ocean and atmosphere will be significantly enhanced during
sea-ice-free conditions.
2. Methods: experimental design
To investigate the response of the atmosphere to projected
changes of sea-ice concentration (SIC), sea-ice thickness
(SIT) and sea-surface temperature (SST), we utilised an
atmospheric general circulation model with the prescribed
boundary conditions at the ocean surface. The atmospheric
model (MGO-3, T63L25) was developed at Voeikov Main
Geophysical Observatory. A short outline of parameterisation of physical processes used in the model, and its use in
weather prediction is provided in the Supplementary File.
Boundary conditions at the ocean surface were selected from
the Coupled Model Intercomparison Project (CMIP5),
which runs with the requirements that the coupled model
should properly simulate the annual cycle of Arctic SIC,
SIT and SST for the 19792012 period (Supplementary
Figs. 13). In our study, time-varying monthly SIC, SIT and
SST were selected from the Max Plank Institute for
Meteorology Earth System Model (MPI ESM MR), which
satisfies these requirements.
According to the MPI ESM MR climate simulations with
forcing scenario RCP8.5, the Arctic SIC will persistently
be absent in September 20702079, which was therefore
designated as the sea-ice-free period. Three sets of ensemble
simulations with the atmospheric model were conducted
using different surface ocean boundary conditions: (1)
Climate simulation for the control period (19801989)
with prescribed time-varying distribution of SIC, SIT and
SST and observed CO2 concentration in the atmosphere,
(2) Climate simulation of the sea-ice-free period (20702079)
with prescribed time-varying distribution of SIC, SIT
and SST using forcing scenario RCP8.5, and (3) Climate
simulation with prescribed time-varying SST in extra-polar
oceans in the control period (19801989) and time-varying
SIC, SIT and SST in the polar region for the sea-ice-free
period (20702079). It is important to note that SIC/SIT/
SST changes in the polar region are integrated parts of
global climate warming. Therefore, the purpose of the third
climate simulation is to quantify the response of the atmosphere to regional forcing that we consider as analogous to
Arctic amplification forcing.
The ensemble of experiments started from a random
initial state of the atmosphere and following initialisation
of the atmosphere to the prescribed boundary condition. The
difference between the sea-ice-free and control climates will
be termed as climate change due to global forcing, or GLB
simulation. Differences between simulation with SIC/SIT/
SST anomalies in the polar region and simulation of control
climate will be termed as climate change due to polar forcing,
or POL simulation that represents the Arctic amplification.
In order to detect and quantify the potential response of
the atmospheric variability to different boundary conditions
that may be masked by large internal variability in high
latitudes, we conducted 30 experiments of 10-yr duration
for all three sets of simulation, thus providing 300 samples
for analysis of monthly and seasonal changes. Statistical
significance of the atmospheric response was estimated using
unpaired t-test with level of confidence 95% (a0.05).
3. Arctic amplification and its impact on the
polar atmosphere
Observational studies have indicated that annual nearsurface air temperature poleward 608N increased at twice
the rate of the Northern Hemisphere warming during
19792008 (Bekryaev et al., 2010) and further increased
Arctic amplification is evident in the most recent data
analysed through 2014 (Johannessen et al., 2016). Arctic
amplification is apparent throughout the year, but is
manifested most strongly in autumn and winter. The spatial
structure of Arctic amplification due to sea-ice loss is
strongly determined by the specific structure of sea-ice loss
pattern (Kumar et al., 2010; Screen and Simmonds, 2010).
The questions to be answered are whether Arctic amplification will change with projected sea-ice loss during the
21st century and whether it depends on the strength of
anthropogenic forcing scenario. We find that the seasonal
ARCTIC AMPLIFICATION AND THE JET STREAM
and annual values of Arctic amplification do not show
notable dependence on the projected sea-ice extent at least
for the 21st century (Supplementary Fig. 4).
In the GLB simulations, the annual global warming for
the sea-ice-free period with respect to the control run is
3.0 8C, moderately greater in winter (3.2 8C) and less in
summer (2.7 8C). Annual warming over the Arctic Ocean
and adjacent land (60908N) amounts to 6.6 8C with the
largest warming in autumn (7.0 8C) and winter (10.7 8C)
(Fig. 1a). This implies that the annual Arctic amplification,
here defined as the ratio of the surface air temperature
trend over the polar region (60908N) to global trend, is
2.2. In autumn, amplification is close to the annual value
2.3, and in winter, amplification is usually much stronger
than the annual value 3.4 (Supplementary Fig. 2).
In the POL simulation with prescribed SIC/SIT/SST
anomalies in the Arctic Ocean, annual warming over the
polar region (60908N) is 4.5 8C, which is just 68% as
large as the warming over the same region in the GLB
simulation. Although the size and strength of prescribed
SIC/SIT/SST anomalies are identical in both the GLB and
POL simulations, the response of the surface air temperature in the polar region appears essentially weaker in the
POL simulation. This is because the run with global forcing
enhances northward heat transport of the global atmosphere, and it provides additional near-surface warming up
to 2.1 8C (32%) over the polar region. The warming due
to the SIC/SIT/SST anomalies depends on season and
amounts to 50% in autumn and 75% in winter relative to
the warming in case of global forcing (Fig. 1). Our finding
is in qualitative agreement with a similar assessment that
was conducted using two atmospheric models forced with
the observed evolution of SIC and SST for the period
19792008 (Screen et al., 2012). The study accounts for
roughly 70% of the simulated annual near-surface warming
in the region northward 678N, indicating that the proportion of POL to GBL forcing will be about the same in
projected and present climates.
3
Evidence from observations indicates that total northward heat transport occurs in the whole troposphere throughout the year, and it is most pronounced in the cold season
(Peixoto and Oort, 1992). The total heat transport comprises sensible and latent heat, and the latent heat is more
than an order of magnitude less than sensible heat. For
comparison purposes, we evaluate meridional heat transports in the control run (Fig. 2). Computation shows that
in the mid-troposphere, the total northward heat transport across 608N latitude amounts to about 4.2 1014 W
during autumn and winter. In the lower troposphere,
northward heat transport appears to be larger, and
it amounts to 6.2 1014 W and 1.0 1015 W in autumn
and winter, respectively. In the control period, about
90% of total heat is transported across 608N by stationary
eddies (Fig. 2c) and transient eddies (Fig. 2b), and the
remainder by mean meridional circulation (Fig. 2d).
Furthermore, heat transport by transient eddies exceeds
transport by stationary eddies by a factor of two over the
whole troposphere in autumn, while both meridional fluxes
have about the similar strength in winter. In the GLB
simulation (Fig. 3, left), the change of northward transport
of sensible heat shows that its largest increase takes place in
the mid-troposphere centred at 500 hPa in autumn and
winter (Fig. 3a). In contrast, strong southward increase of
heat transport was also originated in the lower troposphere
from the Arctic Ocean causing a reduction of northward
heat transport such a finding has not been reported before.
To get a better understanding of the physical mechanisms
involved in enhancement of the zonally averaged meridional
transport of sensible heat, we decomposed the change of its
total transport into the contribution of three components
(Peixoto and Oort, 1992): heat transport by transient eddies
(Fig. 3b), large-scale stationary eddies (Fig. 3c) and mean
meridional circulation (Fig. 3d). Further, we computed
zonally averaged heat transport across 608N latitude of
200 hPa deep-layer centred at 500 and at 850 hPa, which
we consider as characteristic meridional heat fluxes in the
Fig. 1. Surface warming in polar region due to different climate forcings. Change of surface air temperature (8C) in autumn and winter
for a sea-ice-free period relative to control climate computed with (a) global (GLB) and (b) polar forcing (POL). Stippled areas represent
statistically significant changes with confidence level 95%.
4
V. P. MELESHKO ET AL.
Fig. 2. Meridional heat transport by the atmosphere in control-run climate. Seasonal zonally averaged transport of sensible heat
(8C ×m ×s 1) from 40 to 908N computed in an atmospheric model with prescribed SIC, SIT and SST for the control period (19801989).
Here the total heat transport (a) is decomposed into transport by transient eddies (b), large-scale stationary eddies (c) and mean meridional
circulation (d). Northward transport is positive, and southward is negative. Stippled areas indicate that the absolute value of the transport
is larger than its SD.
mid and lower troposphere. In the following, the heat
transport change within these limits implies its value relative
to that in the control climate.
In the case of the GLB forcing, total northward heat
transport across 608N latitude in autumn increases by
0.4 1014 W (10%) in the mid-troposphere compared to
the control climate (Fig. 3a). Two-thirds of the transport
comes from stationary eddies and one-third from transient
eddies. In winter, the total northward heat transport increases by about 0.9 1014 W (22%), and it is mainly due
to the impact of stationary eddies (Fig. 3c). In contrast,
total northward heat transport in the lower troposphere
across 608N latitude decreases by about 0.7 1014 W (12%)
in autumn and by 0.6 1014 W (6%) in winter compared
to the control climate. This is due to warming of the Arctic
Ocean and enhancement of southward heat transport from
the warmer Arctic Ocean to the continents. The main contribution to the decrease of northward heat transport is
provided by transient eddies (Fig. 3b). In the POL simulation (Fig. 3, right), there is no increase of northward heat
transport in the mid-troposphere. However, enhancement
of the southward heat transport in the lower troposphere
associated with transient eddies remains about the same
strength in both the GLB and POL simulations, thereby
indicating that the main driver changing the heat transport
is Arctic amplification.
In the GLB simulation (Fig. 4a, left), large warming
occurs from the surface boundary layer up to 500600 hPa
in autumn and winter. A significant increase of air temperature (T) also occurs in the whole troposphere throughout the year. However, the POL simulation (Fig. 4a, right)
shows weaker and shallow warming over the polar region.
Its vertical propagation is restricted to the lower troposphere, and this is due to no enhanced northward heat
transport in the troposphere.
Several diagnostic studies have indicated that the Geopotential Height (GPH) has increased with height in the
polar troposphere in recent decades (Overland and Wang,
2010; Francis and Vavrus, 2012) and that this has been
attributed to Arctic amplification. Large GPH increase
with height caused by the decrease of the northward
temperature gradient in the troposphere is also produced
ARCTIC AMPLIFICATION AND THE JET STREAM
5
Fig. 3. Enhancement of northward heat transport. Change of zonally averaged meridional transport of sensible heat (8C×m×s 1) in
autumn and winter from 40 to 908N during the sea-ice-free period relative to control climate computed with global (GLB) and polar (POL)
forcings. The total heat transport (a) is decomposed into transport by transient eddies (b), large-scale stationary eddies (c) and the mean
meridional circulation (d). Northward transport is positive and southward is negative. Stippled areas represent statistically significant
changes with confidence level 95%. Note that the colour scale is different than in Fig. 2.
in the GLB simulation (Fig. 4b, left) in high latitudes.
But when only Arctic amplification forcing is considered in
the POL simulation, the GPH increase is very small and
nearly constant with height across the troposphere, in
contrast with the GLB simulation. For example, changes
of 5001000 hPa thickness in the GLB (POL) simulation
are 7899 m (2097 m) in autumn and 82917 m (23911 m)
in winter. Our results are consistent with recent studies
(Deser et al., 2015; Perlwitz et al., 2015) in which impact of
observed and projected sea ice loss on temperature and
GPH in the low and upper polar troposphere was also
evaluated. However, our study shows more pronounced
quantitative impact since evaluation was made for seasons
with largest response during summer sea-ice-free period.
Therefore, we conclude from Figs. 3 and 4 that the increase
of GPH with height is associated mainly with enhancement
of northward heat transport in the mid-troposphere, and
that the main mechanism providing enhancement of this
transport is global warming, not Arctic amplification.
Arctic amplification provides only a minor contribution
to the increase of GPH and decrease of air temperature
gradient in the mid-troposphere. This is because the
amplification is not strong enough, as its forcing is confined
to a shallow layer of the lower troposphere.
We also compared changes of zonally averaged air
temperature gradient at 850 and 300 hPa in the latitudinal
zone 50808N from the GLB and POL simulations. It is
apparent that the decrease of temperature gradient at 300 hPa
is several times larger in GLB while its decrease at 850 hPa
does not differ substantially in both simulations. Our results
disagree with the earlier suggested hypothesis (Francis
and Vavrus, 2012, 2015) claiming that the Arctic amplification plays the dominant role in the decrease of northward temperature gradient and the increase of GPH with
height in the polar troposphere thereby impacting the polar
jet stream.
4. Temperature variability in the Northern
Hemisphere for an ice-free Arctic Ocean
Our atmospheric model can capture the large-scale patterns
of observed T and GPH subseasonal variability in the
Northern Hemisphere, although their strength appears to
6
V. P. MELESHKO ET AL.
Fig. 4. Changes in the polar troposphere due to different climate forcings. Change of zonally averaged: (a) air temperature (8C) and (b)
geopotential height (m) in autumn and winter from 40 to 908N during the sea-ice-free period relative to control climate computed with
global (GLB) and polar (POL) forcings. Stippled areas represent statistically significant changes with confidence level 95%.
be underestimated compared with observations (Kiktev
et al., 2011) a problem that is common to many climate
models and depends mainly on model resolution. However,
we find it useful to examine how subseasonal variability of
T and GPH will change for a sea-ice-free period and what
impact it might have on atmospheric circulation. Subseasonal variability is determined as root mean square deviation of daily T and GPH from daily multiyear means.
Monthly variability was computed from above mentioned
three sets of experiments, and seasonal changes were defined from monthly distributions. We expect that the size
of the ensemble experiments is large enough to detect
robust response of the atmospheric circulation variability
against background natural variability in high latitudes.
For example, it was shown that an ensemble size of at least
about 50 members is desirable to detect a significant thermodynamical response of atmospheric circulation to Arctic sea
ice loss (Screen et al., 2014), and we used 300 samples in
assessment of response.
The GLB and POL simulations show a statistically significant decrease of T variability in the cold seasons in both
the high- and mid-latitudes (Fig. 5a). The patterns are similar
in both the simulations, and its magnitude is largest over the
Arctic Ocean. Vertical propagation of T variability is very
similar to the extent of the warming in low troposphere in
the case of the POL simulation (Figs. 4a and 5b). This phenomenon is strongly associated with the influence of ocean.
When the ice-free area expands, the Arctic Ocean begins
to play a role of a large polar thermostat. Due to the ocean’s
enormous heat capacity and strong heat exchange with the
atmosphere in cold seasons, air masses in the lower troposphere are well mixed and become thermally more homogeneous in space and time. The area of reduced T variability
further expands to continents due to advection. The most
pronounced decrease of near-surface air temperature variability occurs over area 60908N in autumn (46%) and
in winter (24%) compared to that in control run when
most of the Arctic Ocean is exposed to strong heat interaction with the lower troposphere. A larger decrease of T
variability occurs in autumn because a larger area of open
ocean is exposed to the atmosphere, and this is in qualitative
agreement with similar results obtained from observations
and recent simulation (Screen, 2014; Blackport and Kushner, 2016). In the GLB simulation, a statistically significant increase of T variability is also observed in the
mid-troposphere throughout the year, which apparently is
associated with the enhancement of northward heat transport by large-scale eddies, as indicated before. Furthermore,
some increase in variability of T can also be generated in the
lower troposphere due to natural subseasonal and interannual variability of sea-ice extent in the Arctic Ocean and
adjacent seas. Thus, there are two manifestations of temperature changes in the lower troposphere over the Arctic
Ocean and adjacent continents in cold seasons: (1) warming
amplification, and (2) variability decrease, which does not
depend on Arctic amplification, but strongly depends on the
area of open water in the Arctic Ocean and the season.
ARCTIC AMPLIFICATION AND THE JET STREAM
7
Fig. 5. Change of air temperature in the polar atmosphere. Change of subseasonal variability of: (a) near-surface air temperature (8C) and
(b) seasonally averaged air temperature from 40 to 908N. Both plots represent the sea-ice-free period relative to control climate in autumn
and winter, computed with GLB and POL forcing. Stippled areas represent statistically significant changes with confidence level 95%.
5. Planetary-scale wave changes for an ice-free
Arctic Ocean
To detect and quantify the response of atmospheric circulation patterns including planetary-scale waves to global
and polar forcing, we examined the spatial distribution of
GPH variability change in the troposphere on the assumption that it generally represents statistics about position
and evolution of large-scale oscillation of quasi-stationary
eddies and planetary-scale waves, including statistics of
cyclone tracks. When regional GPH variability increases
in the troposphere, the oscillation of planetary waves and
the occurrence of moving cyclones will increase. However,
when GPH variability decreases, quasi-stationary circulation patterns reinforce and this might promote initiation of
anomalous weather regimes in winter.
In the GLB simulation, large GPH variability develops in
the 200400 hPa (where the largest decrease of temperature
gradient occurs) around the pole mainly in high and midlatitudes in autumn (Fig. 7). In winter, a well-shaped belt
of high variability propagates downward to 500 hPa where
its strength gradually decays with appropriate decrease of
temperature gradient change (Fig. 4). On the other hand, an
extensive pattern with lower GPH variability develops over
Eurasia, the western part of Arctic Ocean and Greenland
in winter, and its strength is most pronounced at 500 hPa.
This implies that depending on the position and intensity of
particular circulation patterns, a decrease of GPH variability
will favour reinforcement of anticyclonic pattern, such as
in Northern Eurasia, or decay of cyclonic activity. Lastly, a
pervasive pattern of increased GPH variability also forms at
850 hPa over the polar region and partly mid-latitudes. Such
an increase appears to be supported with the enhancement of
northward heat transport by large-scale stationary eddies
and mean meridional circulation in the lower troposphere
(Fig. 3). It also appears that in case of GLB simulation
enhancement of northward heat transport produces larger
influence on GPH variability than decrease of temperature
variability in the lower troposphere.
In the GLB simulation, the zonally averaged zonal wind
decreases in the latitudinal belt of 60808N throughout
the troposphere with a maximum in the mid-troposphere
(Fig. 6). The zonal wind decrease is statistically significant
and is qualitatively consistent with some recent observational and simulation studies using atmospheric and coupled
models (Francis and Vavrus, 2012; Deser et al., 2015;
Blackport and Kushner, 2016). As noted when considering
GPH change, the response of zonal wind also turns out to
be more pronounced in our study than in Deser et al. (2015)
and Blackport and Kushner (2016). The larger decrease of
zonal wind occurs in winter and expands further northward
in the upper troposphere thereby indicating that the polar
vortex might become weaker.
In the POL simulation, the response of GPH variability to
SIC/SST anomaly forcing appears to be entirely different
(Fig. 7). Oscillation of well-marked planetary waves does not
develop in the troposphere, and GPH variability decreases
over the whole polar region (60908N) and northern Eurasia
in winter. It is important to note that the SIC/SST anomaly
in the Arctic Ocean and associated decrease of temperature
8
V. P. MELESHKO ET AL.
Fig. 6. Zonal-wind evolution. Change of zonally averaged zonal wind (m×s 1) in autumn and winter during the sea-ice-free
period relative to control climate computed with global and polar forcings. Contours indicate the zonal wind (contour interval of 5 m×s 1,
zero contour thickened) distributions from the control run. Stippled areas indicate statistically significant changes with confidence
level 95%.
variability in the lower troposphere meaningfully contribute
to low GPH variability in the upper and mid-troposphere,
which in turn might favour amplification of the polar vortex.
Statistically significant weakening of zonal wind is also
observed in the POL simulation but to a lesser extent than
in the GLB simulation. This implies that such a decrease of
zonal wind is presumably associated with a decrease of
northward temperature gradient in the troposphere due to
sea ice reduction, but it is not adequate to initiate planetary
wave oscillation.
To further quantify regional/height changes of GPH variability and compare them with reference variability in the
Fig. 7. Oscillation of planetary waves and climate forcing. Change of subseasonal GPH variability (m) at three levels of the troposphere
in autumn and winter during the sea-ice-free period relative to control climate computed with global (GLB) and polar (POL) forcing.
Stippled areas represent statistically significant changes with confidence level 95%.
ARCTIC AMPLIFICATION AND THE JET STREAM
control period, we introduce a GPH variability index (GVI).
The index was computed at 300, 500 and 850 hPa surfaces
in polar region (60908N) and mid-latitudes (40608N) for
domains with a statistically significant increase (GVI)
and decrease (GVI ) of GPH variability (Supplementary
Figs. 6 and 7). GVI and GVI describe the integrated
enhancement or decay of GPH variability (in percent) in
circulation patterns relative to reference variability in polar
region (60908N) and mid-latitudes (40608N). Seasonal
assessment of GVI in the climate simulations using the
GLB and POL forcings indicates the following:
(1) In the global warming, enhancement of variability
occurred in the upper and lower polar troposphere
(60908N), and GVI+ amounts to 9% at 850 hPa
surface in autumn. In part this was because internal
GPH variability was about half as much of variability
in the upper troposphere in cold seasons. It also
appears that the decrease of temperature variability
did not have a marked impact on GPH variability in
the lower troposphere. On the other hand, the major
contribution to negative value of GVI in the midtroposphere was made by the large-scale pattern
of decreased variability extending across Northern
Eurasia to Greenland.
(2) In the GLB warming, prominent growth of variability
(up to 7%) occurred throughout the troposphere in
winter, due to enhancement of the planetary wave
oscillation in the mid-latitudes (40608N).
(3) In case of the polar anomaly forcing, GPH variability
reduced primarily in the polar and mid-latitude
troposphere, and its largest decrease (up to 10%)
took place in the polar troposphere during winter.
Analysis of observation and model simulations has revealed
that northward temperature gradient decreases and jet
flow weakens in the polar troposphere due to global climate
warming. These interdependent phenomena are regarded
as robust features of the climate system. An increase of
planetary wave oscillation that is attributed to Arctic
amplification (Francis and Vavrus, 2012; Francis and
Vavrus, 2015) has not been confirmed from analysis of
observation (Barnes, 2013; Screen and Simmonds, 2013) or
in our analysis of model simulations of projected climate.
However, we found that GPH variability associated with
planetary wave oscillation increases in the background of
weakening of zonal flow during the sea-ice-free summer.
Enhancement of northward heat transport in the troposphere was shown to be the main factor responsible for
decrease of northward temperature gradient and weakening
of the jet stream in autumn and winter. Arctic amplification
provides only minor contribution to the evolution of zonal
flow and planetary wave oscillation. In the meantime, the
9
mechanisms responsible for the increase of wave oscillation
under weakening of zonal flow remain unclear at present,
and this is a challenge for further study.
6. Conclusions
In conclusion, our study did not detect an impact of the
inherent Arctic amplification on planetary wave oscillation
in the troposphere of the mid-latitudes during the cold
seasons. Rather, the oscillation of planetary waves developed and became a well-organised belt in upper and midtroposphere mainly by the influence of global warming
which was the major driver for enhancement of the northward heat transport, decrease of meridional temperature
gradient and subsequent increase of planetary wave oscillation in the upper troposphere. However, because meridional
temperature gradient growth decreased from upper troposphere downward, the oscillation belt did not propagate
below 500 hPa, and it appeared as disintegrated patterns
of less strength. In other words, we also did not reveal any
organised oscillation of planetary waves in the lower troposphere in autumn and winter due to Arctic amplification.
Within this framework, it is worth mentioning an early
study (Newson, 1973) in which response of the atmosphere
to prescribed sea-ice-free anomaly was considered in the
Arctic Ocean. The open water was maintained at freezing
temperature during a persistent winter. Atmospheric simulation with such strong boundary forcing revealed that,
apart from a significant increase of near-surface temperature
over the Arctic Ocean, extensive cooling areas developed
stretching across the whole Eurasia and North America.
These looked like extreme weather patterns occurring occasionally in the mid-latitudes. However, we believe that experimental design and associated boundary forcing from the
Arctic Ocean appear to be far from reality.
In our study, we have quantified the relative contribution
of Arctic amplification and northward heat transport in
the troposphere to changes of the zonally averaged meridional temperature gradient and oscillation of planetaryscale waves in a projected summer sea-ice- free climate. It
has been shown that northward heat transport is the major
factor in decreasing the northward temperature gradient in
the polar atmosphere and increasing the planetary-scale
wave oscillation in the troposphere of the mid-latitudes.
Arctic amplification does not show any essential impact on
planetary-scale oscillation in the mid and upper troposphere, although it does cause a decrease of northward heat
transport in the lower troposphere. These results confound
the interpretation of the short observational record that has
suggested a causal link between recent Arctic melting and
extreme weather in the mid-latitudes.
10
V. P. MELESHKO ET AL.
7. Acknowledgements
This work has been supported by the joint Russian
Norwegian projects NORRUS-CLIMARC (225032) and
RFBR (12-05-93092 Norv_a). The research is also funded
by project RFBR (15-06- 08163 a). We acknowledge the
World Climate Research Programme’s Working Group on
Coupled Modelling, which is responsible for CMIP, and we
thank the climate modelling groups for producing and
making available their model output. For CMIP the U.S.
Department of Energy’s Program for Climate Model
Diagnosis and Intercomparison provides coordinating
support and led development of software infrastructure in
partnership with the Global Organization for Earth System
Science Portals. We also thank Dr. Martin Miles for editing
the article.
References
Barnes, E. A. 2013. Revisiting the evidence linking Arctic amplification to extreme weather in midlatitudes. Geophys. Res. Lett. 40,
47344739. DOI: http://dx.doi.org/10.1002/grl.50880
Bekryaev, R. V., Polyakov, I. V. and Alexeev, V. A. 2010. Role
of polar amplification in long-term surface air temperature
variation and modern arctic warming. J. Clim. 23, 38883906.
Blackport, R. and Kushner, P. J. 2016. The transient and equilibrium climate response to rapid summertime sea ice loss in
CCSM4. J. Clim. 29, 401417.
Budikova, D. 2009. Role of Arctic sea ice in global atmospheric
circulation: a review. Global Planet, Change 68, 149163.
Cohen, J., Screen, J. A., Furtado, J. C., Barlow, M., Whittleston, D.
and co-authors. 2014. Recent Arctic amplification and extreme
mid-latitude weather. Nat. Geosci. 7, 627637.
Comiso, J. C. 2012. Large decadal decline of the Arctic multiyear
ice cover. J. Clim. 25, 11761193.
Deser, C., Tomas, R. A. and Sun, L. 2015. The role of ocean
atmosphere coupling in the zonal-mean atmospheric response to
Arctic sea ice loss. J. Clim. 28, 21682186.
Francis, J. A. and Vavrus, S. J. 2012. Evidence linking Arctic
amplification to extreme weather in mid-latitudes. Geophys. Res.
Lett. 39, L06801.
Francis, J. A. and Vavrus, S. J. 2015. Evidence for a wavier jet
stream in response to rapid Arctic warming. Environ. Res. Lett.
10, 014005.
Honda, M., Inue, J. and Yamane, S. 2009. Influence of low Arctic
sea-ice minima on anomalously cold Eurasian winters. Geophys.
Res. Lett. 36, L08707.
Johannessen, O. M., Bengtsson, L., Miles, M. W., Kuzmina, S.,
Semenov, V. A. and co-authors. 2004. Arctic climate change:
observed and modeled temperature and sea-ice variability.
Tellus A 56, 328341.
Johannessen, O. M., Kuzmina, S., Bobylev, L. P. and Miles, M. W.
2016. Surface air temperature variability and trends in the Arctic:
new amplification assessment and regionalization. Tellus A 68,
28234.
Kattsov, V. M., Ryabinin, V. E., Overland, J. E., Serreze, M. C.,
Visbeck, M. and co-authors. 2010. Arctic sea-ice change: a
grand challenge of climate science. J. Glaciol. 56, 11151121.
Kiktev, D. B., Tolstyh, M. A., Zaripov, R. B., Zaichenko, M. Y.,
Muravyev, A. V. and co-authors. 2011. Joint test of seasonal
prediction technology using global atmospheric models of
Russian Hydrometcenter and MGO. Digest on Testing of
New Technologies of Hydrometeorological Forecasts 38, 413
(In Russian, Online at: http://elibrary.ru/item.asp?id17258831;
http://method.meteorf.ru/publ/sb/sb38/kiktev.pdf).
Kumar, A., Perlwitz, J., Eischeid, J. Quan, X., Xu, T. and
co-authors. 2010. Contribution of sea ice loss to Arctic amplification.
Geophys. Res. Lett. 37, L211701.
Mahlstein, I. and Knutti, R. 2012. September Arctic sea ice
predicted to disappear near 2 8C global warming above present.
J. Geophys. Res. 117, D06104.
Mori, M., Watanabe, M., Shiogama, H., Inoue, J. and Kimoto, M.
2014. Robust Arctic sea-ice influence on the frequent Eurasian
cold winters in past decades. Nat. Geosci. 7, 869873.
Nansen Center Sea Ice Information System, Bergen, Norway.
Online at: http://www.arctic-roos.org
Newson, R. L. 1973. Response of a general circulation model
of the atmosphere to removal of the Arctic ice-cap. Nature 241,
3940.
Overland, J. E., Francis, J., Hall, R., Hanna, E., Ki, S.-J. and
co-authors. 2015. The melting Arctic and mid-latitude weather
patterns: are they connected? J. Clim. 28, 79177932.
Overland, J. E. and Wang, M. 2010. Large-scale atmospheric
circulation changes are associated with the recent loss of Arctic
sea ice. Tellus A 62, 19.
Overland, J. E. and Wang, M. 2013. When will the summer Arctic
be nearly sea ice free? Geophys. Res. Lett. 40, 20972101. DOI:
http://dx.doi.org/10.1002/grl.50316
Peixoto, J. P. and Oort, A. H. 1992. Physics of Climate. New York,
American Institute of Physics.
Perlwitz, J., Hoerling, M. and Dole, R. 2015. Arctic tropospheric
warming: causes and linkages to lower latitudes. J. Clim. 28,
21542167.
Screen, J. A. 2014. Arctic amplification decreases temperature
variance in northern mid- to high-latitudes. Nat. Clim. Change 4,
577582.
Screen, J. A., Deser, C. and Simmonds, I. 2012. Local and remote
controls on observed Arctic warming. Geophys. Res. Lett. 39,
L10709.
Screen, J. A., Deser, C., Simmonds, I. and Tomas, R. 2014. Atmospheric impacts of Arctic sea-ice loss, 19792009: separating
forced change from atmospheric internal variability. Clim. Dyn. 43,
333344.
Screen, J. A. and Francis, J. A. 2016. Contribution of sea-ice loss
to Arctic amplification is regulated by Pacific Ocean decadal
variability. Nat. Clim. Change 6, 856860. DOI: http://dx.doi.
org/10.1038/nclimate3011
ARCTIC AMPLIFICATION AND THE JET STREAM
Screen, J. A. and Simmonds, I. 2010. Increasing fallwinter energy
loss from the Arctic Ocean and its role in Arctic amplification.
Geophys. Res. Lett. 37, L16707.
Screen, J. A. and Simmonds, I. 2013. Exploring links between
Arctic amplification and mid-latitude weather. Geophys. Res.
Lett. 40, 959964.
Snape, T. J. and Forster, P. M. 2014. Decline of Arctic sea ice:
evaluation and weighting of CMIP5 projection. J. Geophys. Res.
119, 546554.
11
Stroeve, J., Holland, M., Kay, J., Maslanik, J. and Barrett, A.
2011. The Arctic’s rapidly shrinking sea ice cover: a research
synthesis. Clim. Change 110, 10051027.
Vihma, T. 2014. Effects of arctic sea ice decline on weather and
climate: a review. Surv. Geophys 35, 11751214.
Supplementary Materials:
Arctic amplification: Does it impact
the polar jet stream?
By VALENTIN P. MELESHKO1*, OLA M. JOHANNESSEN2,3, ANDREY V. BAIDIN1,
TATIANA V. PAVLOVA1, VERONIKA A. GOVORKOVA1, 1Voeikov Main Geophysical
Observatory, St. Petersburg, Russia. 2Nansen Environmental and Remote Sensing Center, N-5006
Bergen, Norway. 3Nansen Scientific Society, N-5006 Bergen, Norway.
*Correspondence to: [email protected]
This file includes:
Supplementary text
Figs S1 to S7
References
1. A brief description of the atmospheric model
The atmospheric general circulation model (MGO-3) was developed at the Voeikov Main
Geophysical Observatory. The model has T63 spectral resolution corresponding to a grid size
of about 190 km and 25 vertical levels (T63L25). Detailed description of numerical methods
and parameterization of physical processes used in the model can be found in a separate
publication (Meleshko et al., 2014). The model incorporates the major physical processes
governing the atmospheric circulation: solar and terrestrial radiation transfer, convection, largescale and convective precipitation, clouds, and turbulent fluxes of sensible/latent heat in the
boundary layer and in the upper atmosphere. The radiation package accounts for diurnal and
annual variation of solar fluxes at the top of the atmosphere and includes greenhouse gases and
aerosols. In calculating surface fluxes, a grid box considers three types of surfaces with
different physical properties (water, sea ice and land). The atmospheric model is coupled to a
land-surface block that takes into account processes over vegetation of different types, bare soil
and snow-covered surface. The land-surface model describes heat/water transfer and water
discharge in the ground of 3-m depth that is represented with 4 layers. The model has been
validated and it realistically reproduces the main large-scale patterns of the atmospheric
circulation through the seasonal cycle (Kiktev et al., 2011; http://seakc.meteoinfo.ru).
The atmospheric model participates in the Russian national program for monthly to seasonal
weather prediction (Kiktev et al., 2011), and its output products are deployed on the internet
site of the Eurasian climate center (http://seakc.meteoinfo.ru). Seasonal forecasts are also
forwarded to the APEC (Asia-Pacific Economic Cooperation) Climate Centre (APEC Climate
Center, 2006) on a regular basis and are utilized there for multi-model weather prediction on up
to seasonal timescales.
2. Selection of SIC, SIT and SST boundary conditions from CMIP5 simulations
Boundary conditions at the ocean surface were taken from CMIP5 models that proved
successful in simulation of SIC in the Arctic and the annual cycle of SST in the Northern
Hemisphere during 1979–2012. Another requirement for selection of boundary condition was
that the climate models had to simulate a persistent sea-ice-free period of at least 10-years
duration in the 21st century with forcing scenario RCP8.5. According to the employed metric,
CMIP5 models should also realistically reproduce the observed multiyear trend in sea ice and
sea-ice extent in September and March.
For analysis of the sea-ice distribution simulated by 39 CMIP5 models, we used the
HadISST (Rayner et al., 2003, updated) data set, and the annual cycle of sea ice was validated
using a Taylor diagram (Taylor, 2001). The validation showed that 10 climate models
reproduced the observed sea-ice extent within an acceptable bias of 1 × 106 km2 in September
and March 1979–2012 (Pavlova and Kattsov, 2013). The success rate in simulating SST annual
cycle was evaluated in terms of cumulative seasonal RMSE over Northern Hemisphere
(Meleshko and Govorkova, 2013). The results showed that the boundary conditions of seven
AOGCMs met the specified criteria. Because some models had common origin in their design,
we selected boundary conditions from three independent models (IPCC, 2013):
1. ACCESS1-0 (CSIRO and Bureau of Meteorology, Australia);
2. CESM1-CAM5 (NSF-DOE-NCAR, USA); and
3. MPI-ESM-MR (Max Planck Institute for Meteorology, Germany).
The negative trend of observed sea-ice extent in September relative to 1981–2010 is 13.3% per
decade (Fetterer et al., 2002, updated), while the three climate models slightly underestimate
the observed trend (11%, 10%, 9%) – see Fig. S1. Regarding the “sea-ice-free” period, the ice
cover is projected to be reduced by 95–100% in August–October 2070–2079 relative to 1980–
1989, and the internal variability of sea ice will not exceed 0.6 × 106 km2 in that period (Fig.
S2). In February–April, when winter sea-ice extent is largest, its reduction is projected to be
14–24%. The monthly mean increase of SST in July–October in the polar region (60º–90ºN) is
significant for the sea-ice-free period (2070–2079) (Fig. S3). However the annual variation of
monthly SST increase will be much less south of 60ºN, with a mean annual SST increase of
~2.5°С.
Three sets of ensemble experiments were carried out to evaluate response of the model
atmosphere to variation in three sets of SIC, SIT and SST boundary conditions. Examination of
results did not reveal any specific distinction in responses of the atmosphere to some
differences in boundary forcing that would be of interest to the study. Therefore we further
considered results of simulation with prescribed boundary condition only from MPI-ESM-MR
model. Monthly SIC, SIT and SST fields from the MPI model grid were interpolated to the
MGO model grid and in the course of running the atmospheric model, daily values of SIC, SIT
and SST were computed and utilized.
2
Fig. S1. Long-term sea-ice decrease in the Arctic. Temporal evolution of September sea-ice extent 1980–
2100 computed by three CMIP5 models (ACCESS1-0, CESM1-CAM5 and MPI-ESM-MR) with forcing
scenario RCP8.5. Sea-ice extent from the HadISST1 observational data set is shown in red (Rayner et al.,
2002, updated).
Fig. S2. Annual cycle of sea ice decrease. Monthly change of sea ice extent (%) in the Arctic Ocean for the
sea-ice-free period (2070–2079) relative to its extent in control climate (1980–1989) derived from three
CMIP5 model simulations with forcing scenario RCP8.5.
Fig. S3. Annual cycle of SST increase. Monthly change of sea-surface temperature (ºC) in the Arctic Ocean
(60º–90ºN) for the sea-ice-free period (2070–2079) relative to temperature in control climate (1980–1989)
derived from three CMIP5 model simulations with forcing scenario RCP8.5.
3
3. Arctic amplification from CMIP5 models
The relative contribution of different physical processes and feedbacks responsible for Arctic
amplification has been widely discussed in numerous studies (Serreze and Francis, 2006;
Graversen and Wang, 2009; Lu and Cai, 2009; Hwang et al., 2011; Serreze et al., 2011;
Graversen et al., 2014; Gao et al., 2015; Johannessen et al., 2016). A number of concepts are
proposed to explain the mechanisms and physical processes responsible for initiation of such a
phenomenon. Among them solar and long-wave fluxes play a significant role through the
change of surface albedo, atmospheric water vapor and cloudiness (Screen and Simmonds,
2012; Cohen et al., 2014; Graversen et al., 2014). A decrease of sea-ice cover in the Arctic and
the much lower albedo of the water surface favors increase of the solar energy absorption by
the upper ocean layer and its temperature increase in summer. However in autumn when air
temperature decreases and becomes lower than ocean temperature, heat accumulated during
summer is transferred back to the atmosphere through radiative emission and by means of
sensible and latent heat exchange and thereby warms the low layer of the atmosphere. The heat
stored in the ocean also causes a delay in building up of ice cover and reduces ice thickness. An
increase of the open water areas contributes to an increase of air temperature and water vapor
in the lower layer of the atmosphere over the Arctic Ocean and the adjacent terrestrial areas.
Apart from local ocean–atmosphere interaction processes, a contribution of about 25% to
the Arctic amplification is also provided by northward transport of heat and water vapor
(Hwang et al., 2011; Screen et al., 2012; Cohen et al., 2014) but it is difficult to evaluate the
anthropogenic component of amplification due to implication of other natural forcings in the
climate system (Johannessen et al., 2004; Suo et al., 2013; Cohen et al., 2014) Therefore we
evaluated the Arctic amplification due to global anthropogenic forcing from CMIP5 climate
simulation for the 21st century. Thirty CMIP5 model runs were selected for the period 1980–
2099 with the forcing scenario RCP8.5. The Arctic amplification for each model was computed
as the ratio of surface air temperature (SAT) trend in the polar region (60°–90ºN) to the global
trend The entire time period (1980–2099) was divided into two sub-periods (1980–2039 and
2040–2099) and appropriate trends and amplification were computed. We found that seasonal
and annual Arctic amplification evaluated from model ensembles using two sub-periods did not
show noticeable dependence on projected sea-ice extent at least for the 21st century (see Fig.
S4). However, the standard deviation of amplification appears to be larger in winter and spring
in the second half of the century, when the sea-ice cover is significantly reduced as compared
to the beginning of the century.
4
Fig. S4. Arctic amplification. Seasonal and annual Arctic amplification and its standard deviation computed
from ensemble of 30 CMIP5 models using forcing scenario RCP8.5 for the periods 1980–2039 and 2040–
2099 with different sea-ice extent in the Arctic Ocean.
The Arctic amplification resulting from global anthropogenic forcing, wherein the ocean
and cryosphere play a significant role in the polar region, probably remains in thermodynamic
quasi-equilibrium state with respect to global climate warming due to radiative forcing. At the
same time, recent studies have shown that strength of Arctic amplification can vary depending
on the phase of the Atlantic Multidecadal Oscillation and Pacific Decadal Oscillation
(Johannessen et al., 2016; Screen and Francis, 2016).
From Fig. S5, it is apparent that some models show a significant deviation of amplification
from mean values for all seasons. It appears that the differences are associated with
parameterization of physical processes and appropriate feedbacks in climate models.
Fig. S5. Seasonal spread of Arctic amplification. Probability distribution of seasonal Arctic amplification
derived from ensemble of 30 CMIP5 climate models with forcing scenario RCP8.5. Arctic amplification is
defined as the ratio of surface air temperature trend over the polar region (60º–90°N) to its global trend. The
linear trends of surface air temperature were computed for two periods of sea-ice extent: 1980–2039 and
2040–2099. The seasonal probability in the colour bars was computed as the ratio of number of samples falling
into an appropriate range to the total 60 samples comprising two periods with 30 members in each period.
5
Fig. S6. GPH Variability Index (GVI, %) for the sea-ice-free period in the polar region (60°–90°N) in autumn
(SON) and winter (DJF). GVI represents integrated change of seasonal GPH variability. Its computation was
made separately for domains with statistically significant positive (GVI+) and negative (GVI–) variability y
differences, multiplied by proportion of the appropriate domain and normalized by mean reference variability
computed over the same domain in the control period. Area with statistically significant changes of both signs
is equal to unity. GVI+ and GVI– are computed at three surfaces (300 hPa, 500 hPa, 850 hPa) in simulations
with GLB and POL forcing.
Fig. S7. Geopotential Height (GPH) Variability Index (GVI, %) for the sea-ice-free period in the mid-latitudes
(40º–60ºN) in autumn (SON) and winter (DJF). GVI represents integrated change of seasonal GPH variability.
Its computation was made separately for domains with statistically significant positive (GVI+) and negative
(GVI–) variability differences, multiplied by the proportion of the appropriate domain and normalized by
mean reference variability computed over the same domain in the control period. Area with statistically
significant changes of both signs is equal to unity. GVI+ and GVI– are computed at three pressure-surface
levels (300 hPa, 500 hPa, 850 hPa) in simulations with GLB and POL forcings.
6
References
APEC Climate Center. 2006. Assessment of the climate forecasts produced by individual models and MME
method. APCC Technical Report 1.
Cohen, J., Screen, J. A., Furtado, J. C., Barlow, M., Whittleston, D. and co-authors. 2014. Recent Arctic
amplification and extreme mid-latitude weather. Nature Geosci. 7, 627-637.
Fetterer, F., Knowles, K., Meier, W. and Savoie, M. 2002, updated daily. Sea Ice Index. NSIDC. Boulder.
Colorado USA.
Gao, Y., Sun, J., Li, F., He, S., Sandven, S. and co-authors. 2015. Arctic sea ice and Eurasian climate: a review.
Adv. Atmos. Sci. 32, 92–114.
Graversen, R. G., Langen, P. L. and Mauritsen, T. 2014. Polar amplification in CCSM4: contributions from the
lapse rate and surface albedo feedbacks. J. Climate 27, 4433–4450.
Graversen, R. G. and Wang, M. 2009. Polar amplification in a coupled climate model with locked albedo. Clim.
Dyn. 33, 629–643.
Hwang, E-T., Frierson, D. M. W. and Kay, J. E. 2011. Coupling between Arctic feedbacks and changes in
poleward energy transport. Geophys. Res. Lett. 38, L17704.
IPCC Fifth Assessment Report Climate Change 2013: The Physical Science Basis (eds T. F. Stocker et.al.) 1535
(Cambridge Univ. Press, 2013).
Johannessen, O. M., Bengtsson, L., Miles, M. W., Kuzmina, S., Semenov, V. A. and coauthors. 2004. Arctic
climate change: observed and modeled temperature and sea-ice variability. Tellus 56A, 328–341.
Johannessen, O. M., Kuzmina, S., Bobylev, L. P. and Miles, M. W. 2016. Surface air temperature variability and
trends in the Arctic: New amplification assessment and regionalization. Tellus 68A, 28234.
Kiktev, D. B., Tolstyh, M. A. Zaripov, R. B., Zaichenko, M. Y., Muravyev, A.V. and co-authors. 2011. Joint test
of seasonal prediction technology using global atmospheric models of Russian Hydrometcenter and MGO.
Digest on testing of new technologies of hydrometeorological forecasts 38, 4–13 (In Russian, available at
http://elibrary.ru/item.asp?id=17258831, http://method.meteorf.ru/publ/sb/sb38/kiktev.pdf).
Lu, J. and Cai, M. 2009. Seasonality of polar surface warming amplification in climate simulations. Gephys. Res.
Lett. 36, L16704.
Meleshko, V. P. and Govorkova, V. A. 2013. Performance of CMIP3 and CMIP5 models in simulation of current
climate. VMGO Proceedings 568, 26–50. (In Russian, available at
http://voeikovmgo.ru/images/stories/publications/568.pdf)
Meleshko, V. P., Matyugin, V. A., Sporyshev, P. V., Pavlova, T. V., Govorkova, V. A. and co-authors. 2014.
General circulation model of the atmosphere (version MGO-03 T63L25). VMGO proceedings. 571, 5–87. (In
Russian. available at http://voeikovmgo.ru/images/stories/publications/2014/571.pdf).
Pavlova, T. V. and Kattsov, V. M. 2013. Sea ice cover of the word ocean computed in CMIP5 models. VMGO
Proceedings 568, 7–25 (In Russian, available at http://voeikovmgo.ru/images/stories/publications/568.pdf).
Rayner, N. A., Parker, D. E., Horton, E. B., Folland, C. K., Alexander, L. V. and co-authors. 2003, updated. Global
analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century.
J. Geophys. Res. 108, 4407.
Screen J. A. and Francis, J. A. 2016. Contribution of sea-ice loss to Arctic amplification is regulated by Pacific
Ocean decadal variability. Nature Climate Change, 6, doi:10.1038/nclimate3011.
Screen, J. A. and Simmonds, I. 2010. Increasing fall-winter energy loss from the Arctic Ocean and its role in
Arctic amplification. Geophys. Res. Lett. 37, L16707.
Serreze, M. C., Barrett, A. and Cassano, J. C. 2011. Circulation and surface controls on the lower tropospheric air
temperature field in the Arctic. J. Geophys. Res. 116, D07104.
Serreze, M. and Francis, J. 2006. The Arctic amplification debate. Clim. Change 76, 241–264.
Suo, L., Ottera, O. H., Bentsen, M., Gao, Y. I. and Johannessen, O. M. 2013. External forcing of the early 20th
century Arctic warming. Tellus 65A, 1–14.
Taylor, K. E. 2001. Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res.
106, 7183–7192.
7