Evidence for a wavier jet stream in response to rapid Arctic warming

Abstract

New metrics and evidence are presented that support a linkage between rapid Arctic warming, relative to Northern hemisphere mid-latitudes, and more frequent high-amplitude (wavy) jet-stream configurations that favor persistent weather patterns. We find robust relationships among seasonal and regional patterns of weaker poleward thickness gradients, weaker zonal upper-level winds, and a more meridional flow direction. These results suggest that as the Arctic continues to warm faster than elsewhere in response to rising greenhouse-gas concentrations, the frequency of extreme weather events caused by persistent jet-stream patterns will increase.

Full text

Environ. Res. Lett. 10 (2015) 014005 doi:10.1088/1748-9326/10/1/014005
LETTER
Evidence for a wavier jet stream in response to rapid Arctic warming
Jennifer A Francis
1
and Stephen J Vavrus
2
1
Institute of Marine and Coastal Sciences, Rutgers University, New Brunswick, New Jersey, USA
2
Center for Climatic Research, University of Wisconsin-Madison, Madison, Wisconsin, USA
E-mail: francis@imcs.rutgers.edu
Keywords: jet stream, Arctic amplification, extreme weather
Abstract
New metrics and evidence are presented that support a linkage between rapid Arctic warming, relative
to Northern hemisphere mid-latitudes, and more frequent high-amplitude (wavy) jet-stream config-
urations that favor persistent weather patterns. We find robust relationships among seasonal and
regional patterns of weaker poleward thickness gradients, weaker zonal upper-level winds, and a more
meridional flow direction. These results suggest that as the Arctic continues to warm faster than else-
where in response to rising greenhouse-gas concentrations, the frequency of extreme weather events
caused by persistent jet-stream patterns will increase.
This paper builds on the proposed linkage between
Arctic amplification (AA)—defined here as the
enhanced sensitivity of Arctic temperature change
relative to mid-latitude regions—and changes in the
large-scale, upper-level flow in mid-latitudes [1,2].
Widespread Arctic change continues to intensify, as
evidenced by continued loss of Arctic sea ice [3];
decreasing mass of Greenland’s ice sheet [4], rapid
decline of snow cover on Northern hemisphere
continents during early summer [5], and the contin-
ued rapid warming of the Arctic relative to mid-
latitudes. While these events are driven by AA, they
also amplify it: melting ice and snow expose the dark
surfaces beneath, which reduces the surface albedo,
further enhances the absorption of insolation, and
exacerbates melting. Expanding ice-free areas in the
Arctic Ocean also lead to additional evaporation that
augments warming and Arctic precipitation [6].
Traditionally AA is measured as the change in sur-
face air temperature in the Arctic relative to either the
Northern hemisphere or the globe [7]. It arises owing
to a variety of factors, including the loss of sea-ice and
snow, increased water vapor, a thinner and more frac-
tured ice cover, and differences between the Arctic and
lower latitudes in the behavior of lapse-rate and radia-
tive feedbacks [8–13]. Here we do not address the rela-
tive importance of various factors causing AA, but it is
clear from the height-latitude anomalies of air tem-
perature, geopotential, and zonal wind (figure 1) that
AA results in large part from near-surface heating,
although contributions from poleward heat transport
may also play a role [14].
Seasonal time series and trends in AA based on two
metrics and varying initial years are presented in
figure 2. The more traditional method of assessing AA
is to subtract changes in near-surface (1000 hPa) air
temperature anomalies in mid-latitudes (60–30°N)
from those in the Arctic (left side of figure 2). A posi-
tive value of AA indicates that the Arctic is warming
faster than mid-latitudes. Both the time series and
progressive 15 year trends (figure 2, bottom) indicate
an increasingly positive AA in all seasons, particularly
in fall and winter, in agreement with previous analyses
[8]. Starting in the 1990s, coincident with an acceler-
ated decline in Arctic sea-ice extent [3], AA values and
trends became positive in all four seasons for the first
time since the beginning of the modern data record in
the late 1940s, illustrating the Arctic’s enhanced sensi-
tivity to global warming.
The right side of figure 2presents an alternative
metric for AA based on the difference in the
1000–500 hPa thickness change in the Arctic relative
to that in mid-latitudes (same zones as for the tradi-
tional method). Arguably the thickness difference is
more relevant for assessing the effects of AA on the
large-scale circulation, as it represents differences in
warming over a deeper layer of the atmosphere that
should more directly influence winds at upper levels.
Several recent autumns have exhibited strong warm-
ing anomalies in some mid-latitude areas, contribut-
ing to the weakened positive trend after 2007. It is
OPEN ACCESS
RECEIVED
4 November 2014
ACCEPTED FOR PUBLICATION
11 December 2014
PUBLISHED
6 January 2015
Content from this work
may be used under the
terms of the Creative
Commons Attribution 3.0
licence.
Any further distribution of
this work must maintain
attribution to the author
(s) and the title of the
work, journal citation and
DOI.
© 2015 IOP Publishing Ltd

important to note the recent emergence of the signal of
AA from the noise of natural variability: since ∼1995
near the surface and since ∼2000 in the lower tropo-
sphere. This short period presents a substantial chal-
lenge to the detection of robust signals of atmospheric
response amid the noise of natural variability [15,16].
Thus for this study we define the period from 1995 to
2013 as the ‘AA era.’While this demarcation is con-
sistent with previous studies [17], we also investigate
the effects of choosing different commencement years
on detecting changes in the frequency of high-ampli-
tude jet-stream configurations.
Figure 1. Annual-mean anomalies in air temperature (left), geopotential (middle), and zonal winds (right) during 1995–2013 relative
to 1981–2010 for 40–80°N and 1000–250 hPa. Data were obtained from the NCEP/NCAR Reanalysis at http://www.esrl.noaa.gov/
psd/.
2
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

The following linkage between AA and mid-lati-
tude weather patterns has been hypothesized [1].
Increasing AA weakens the poleward temperature
gradient—a fundamental driver of zonal winds in
upper levels of the atmosphere—which causes zonal
winds to decrease, following the thermal wind rela-
tionship [18]. A weaker poleward temperature gra-
dient is also a signature of the negative phase of the
so-called Arctic oscillation/Northern annular mode
(AO/NAM), in which weaker zonal winds are asso-
ciated with a tendency for a more meridional flow,
blocking, and a variety of extreme weather events in
much of the extratropics [19]. Disproportionate
Arctic warming and sea-ice loss favor a negative
AO/NAM aloft [1,2,20,21] and a Northward
migration of ridges in the upper-level flow [1], fur-
ther contributing to an increased meridional pat-
tern. As the wave amplitude and/or frequency of
amplified flow regimes increases, the incidence of
blocking becomes more likely [2], which reduces
the Eastward propagation speed of the pattern.
Consequently, the associated weather systems per-
sist longer in a particular area. Extreme weather
events caused by prolonged weather conditions
(such as cold spells, stormy periods, heat waves, and
droughts), therefore, should also become more
likely, as illustrated by recent studies linking these
events to high-amplitude planetary waves [22–24].
Because AA is strongest in fall and winter
(figure 2), the atmospheric response is expected to be
largest and observed first in these seasons. Results cor-
roborate this expectation [1], showing a marked
reduction in the poleward thickness gradient and
weaker zonal winds at 500 hPa during fall (OND) and
winter (JFM) since 1979 over the North America/
North Atlantic study region. Others find statistically
significant decreases in zonal-mean zonal winds in the
fall but not in winter [15].
Marked spatial and seasonal variability in the
changing poleward thickness gradient dictates pat-
terns of change in zonal winds. While hemisphere-
mean, mid-latitude, zonal winds at 500 hPa have
decreased by about 10% since 1979 during fall [23], no
robust hemispheric trends are apparent in other sea-
sons owing to the spatial variability of the AA signal.
The objectives of the present study are to examine
regional and seasonal expressions of AA that produce
changes in poleward thickness gradients, correspond-
ing effects on zonal wind speeds, and the hypothesized
increase in highly amplified jet-stream regimes.
Recent studies have presented a mixed picture regard-
ing this atmospheric response to AA. Some observa-
tional analyses find evidence of increased wave
amplitude in certain locations and seasons, but statis-
tical significance is often lacking [1,15,16], likely
owing to the recent emergence of AA from natural
Figure 2. Arctic amplification seasonal time series (a), (b) and trends (c), (d) based on two metrics: (left) differences in 1000 hPa
temperature anomalies (relative to 1948–2013 mean, °C) between the Arctic (70–90°N) and mid-latitudes (30–60°N); (right)
differences in 1000–500 hPathickness anomalies (percentof meanfor eachzone) betweenthe twozones. Contoured colorsin c,d
indicate moving 15 year trends (C/decade for T1000 and percent of mean per decade) for each season. Thex-axis indicatesthe ending
year of each 15 year period between 1949 and 2013, and asterisks indicate ending year of periods with significant trends (confidence
>95%, assessed using an F-test goodness of fit). Data were obtained from the NCEP/NCAR Reanalysis at http://www.esrl.noaa.gov/
psd/.
3
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

variability. Analyses based on climate model simula-
tions are challenged by the sometimes unrealistic
representations of complex Arctic physics and non-
linear atmospheric dynamics. Nevertheless, they, too,
suggest a more meridional flow (often resembling the
negative phase of the AO/NAM) in response to sea-ice
loss [25], and none suggests that the flow will become
more zonal or that planetary waves will decrease in
amplitude. Measuring changes in the strength of the
zonal wind is straightforward, whereas quantifying the
‘waviness’of the circulation is not. We therefore aim
to shed further light on this critical aspect of the link-
age by using new techniques to measure the waviness
of the upper-level flow, and we also comment on the
results of previous efforts to diagnose changes in wave
amplitude.
Seasons are defined as follows: winter (JFM),
spring (AMJ), summer (JAS), and fall (OND). These
definitions are selected to coincide with the summer
minimum and winter maximum of Arctic sea-ice
extent, as well as the onset times of freeze and melt. All
data are from the NCEP/NCAR Reanalysis (NRA)
[26] obtained at http://www.esrl.noaa.gov/psd/.
Analysis of 500 hPa height contours
A simple new method was introduced to assess the
daily meridional amplitude of waves in the upper-level
flow [1]. A single contour in the 500 hPa height field
was selected based on its climatological position within
the strongest gradient, thus representing the path of
the polar jet stream on any individual day. The
planetary wave locations and shapes depicted by height
fields at 500 hPa and those at typical heights of the jet
stream maximum (∼250 hPa) are very similar. The
selected heights of individual contours vary slightly
with season to match climatological jet-stream loca-
tions: within 50 m of 5600 m during the cool/cold
seasons (JFM, AMJ, OND), and 5700 m during
summer months (JAS). Daily height contours are
subsetted from daily mean 500 hPa fields from the
NRA. Correlations of 500 hPa height anomalies
between NRA and either the NCEP Climate Forecast
System Reanalysis or the European Centre for Med-
ium-Range Forecasts Interim values are over 0.99 in all
seasons (not shown), suggesting that mid-tropo-
spheric height fields in NRA are nearly identical to
those of other reanalyses. Small differences in blocking
statistics among various reanalyses have also been
reported [27].
The selection of particular 500 hPa height con-
tours used for analysis of wave amplitude [1] has been
questioned by assertions that the proper contours to
use should be those exhibiting the greatest degree of
waviness [15]. This study reproduced the increased
wave amplitude in the 5600 m contour from 1979 to
2010, but the same analysis based on the contour iden-
tified as the waviest (5300 m) exhibited no increase in
amplitude. As illustrated in figures 3(a) and (b), how-
ever, the mean latitude of the 5300 m contour during
fall (1980–2011) is nearly 15° of latitude farther North
than the mean latitude of the 5600 m contour. More-
over, its more Northerly position is far from the core
of strongest upper-level winds and thus its location
differs substantially from the path of the jet stream.
Winds well North of the jet are substantially weaker
(figures 3(a) and (c)), consequently is it not surprising
that the flow is wavier. Arguably, the analysis of the
more Northerly 5300 m contour does not capture the
location and evolution of the polar jet stream, while
the 5600 m contour more closely tracks the shape of
planetary waves in the strongest upper-level flow.
Note that the mean latitude of the strongest zonal
500 hPa winds is nearly identical in both the AA era
(figure 3(a)) and in earlier years (figure 3(c)), suggest-
ing that contours used here and previously [1] repre-
sent the jet-stream location throughout the satellite
record (since 1979).
Meridional circulation index (MCI)
A key outstanding question in the proposed linkage
between AA and jet-stream behavior is whether
weakened poleward thickness gradients are causing
the upper-level flow to become wavier. One measure
of flow waviness is the ratio of the meridional (North/
South) wind component to the total wind speed. We
propose a simple metric to assess this characteristic of
the flow: the MCI:
=
+
vv
uv
MCI *,
22
where uand vare the zonal and meridional compo-
nents of the wind. When MCI = 0, the wind is purely
zonal, and when MCI = 1 (−1), the flow is from the
South (North). We note that a more meridional flow
can result from either a stronger vand/or weaker u
wind component through simple vector geometry.
Whatever the cause, an increase in |MCI|indicates a
wind vector aligned more North–South and reflects a
changed flow direction. The speed of the meridional
(v) wind may not change, as it is associated with East–
West temperature gradients, but if the total wind
vector becomes more meridional, then the flow is by
definition ‘wavier’. For example, a Northwesterly wind
could shift to a North–Northwesterly wind solely
through a reduction of the Westerly wind component.
For this analysis, MCI is calculated from daily 500 hPa
wind components between 20°N and 80°N at each
gridpoint in NCEP Reanalysis fields.
Coincident anomalies in thickness, zonal
winds, and MCI
In an effort to assess the effects of AA on waviness of
upper-level winds, we compare coincident seasonal
4
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

anomalies during the AA era relative to the period
from 1981–2010. Anomalies in the 1000–500 hPa
thicknesses are presented in the top panels of figures
4–7. During fall (OND, figure 7(a)), when sea-ice loss
exerts its largest direct impact, the pattern of AA
extends across much of the Central Arctic, while
during spring (AMJ, figure 5(a)) and summer (JAS,
figure 6(a)) the areas of positive thickness differences
occur primarily over high-latitude land, likely in
response to earlier snow melt [5]. In all seasons,
positive thickness differences are evident in the North-
west Atlantic. This substantial regional and seasonal
variability illustrates the challenge in detecting robust
hemispheric-mean atmospheric responses to AA,
resulting in the low statistical significance reported in
some previous studies [15,16,28].
The middle panels of figures 4–7present anoma-
lies in zonal wind speeds at 500 hPa corresponding to
anomalies in the poleward gradient of 1000–500 hPa
thicknesses (top panels). Anomalies in |MCI|are
shown in the bottom panels. Immediately obvious is
the close association between the spatial patterns of
weakened poleward gradients (regions where positive
anomalies occur Northward of weaker or negative
anomalies) and areas where zonal winds are weaker.
During winter and autumn (figures 4and 7)a
broad area of substantially weakened poleward
gradient is evident across much of the Northern hemi-
sphere mid-latitudes, particularly in the N. Atlantic
and Northern Eurasia. These areas are closely matched
by the spatial pattern of slower zonal winds, as would
be expected according to the thermal wind relation-
ship. Widespread positive anomalies in |MCI|also cor-
respond to these regions. Changes in the meridional
wind speed, however, are not correlated with either
the changes in poleward gradient or zonal winds, sug-
gesting that changes in |MCI|arise mostly because of
changes in the zonal wind speed. These findings sup-
port the hypothesis that AA causes a more meridional
character to the upper-level wind flow, but this change
is achieved primarily via a reduction in Westerly winds
rather than through an increase in meridional wind
speeds.
Relationships between these variables on a grid-
point-by-gridpoint basis are illustrated in scatterplots
(figure 8). Gridpoints with weaker (stronger) pole-
ward gradients tend to have larger (smaller) |MCI|
values (red scatter-plots), particularly for gridpoints
with the strongest (top decile) total winds, indicative
of the jet stream. In all seasons, robust relationships
between anomalies in the poleward gradient, zonal
winds, and |MCI|are evident. Moreover, in spring,
summer, and fall, the anomaly in 500 hPa zonal winds
accounts for a much larger fraction of the variance in
Figure 3. Zonal-mean zonal winds for fall (OND) from 1995 to 2013 (a) and from 1979 to 1995 (c). Corresponding zonal-mean
geopotential heights for 1979–2013 are shown in (b). Dotted horizontal lines highlight the 500 hPa level, white dashed vertical lines
indicate the latitude of maximum mean zonal winds at 500 hPa, and yellow dashed lines are the latitude of the waviest 500 hPa
contour, as identified in [15].
5
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

|MCI|than does the anomaly in the meridional wind
component (variance explained by U500 in JFM, AMJ,
JAS, OND = 0.42, 0.33, 0.61, 0.38; by V500 = 0.59,
0.02, 0.07, 0.001), suggesting that weakened zonal
winds due to AA are the main factor driving the more
meridional flow in these seasons. We also note that
correlations between differences in 500 hPa mer-
idional wind speeds with either the differences in pole-
ward thickness gradient or zonal wind speed were
small and insignificant, suggesting that changes in the |
MCI|arise primarily because of changes in zonal wind
speeds.
Extreme wave frequency
One aspect of the proposed linkage [1] that heretofore
has been difficult to assess is whether the amplitude of
planetary waves is increasing in response to strength-
ening AA. An alternative metric that we pursue here is
the frequency of highly amplified jet-stream config-
urations. Single contours of daily mean 500 hPa height
fields are used to identify ‘extreme waves’in the jet
stream. Representative contours (5600 m ± 50 m,
except 5700 m ± 50 m in JAS) are selected to represent
the streamline of the strongest 500 hPa winds as
discussed previously, and we note that the selected
contours shift little in latitude with time (figure 3).
Data are analyzed in various longitude zones to
identify days in which the difference between the
maximum and minimum latitudes (ridges and
troughs) of the contour within a region exceeds 35° of
latitude. The threshold of 35° was selected to achieve a
frequency of approximately 20 days per season
(∼20%). Note that individual high-amplitude events,
Figure 4. Anomalies in winter (JFM) (a) 1000–500 hPa thickness (m), (b) zonal wind at 500 hPa (m s
−1
), and (c) the absolute value of
the MCI during 1995–2013 relative to 1981–2010. Data were obtained from NOAA/ESRL at http://www.esrl.noaa.gov/psd/.
6
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

such as blocks and cut-off lows, often persist for
several days, thus the frequency of events < frequency
of high-amplitude days.
The frequency of occurrence of high-amplitude
days is assessed in each season and for the AA era
(1995–2013) relative to the pre-AA period
(1979–1994). We repeat the analysis using two addi-
tional definitions of the AA era—1990–2013 and
2000–2013—to determine the sensitivity of differ-
ences in high-amplitude days to the time period selec-
ted for the AA era. The mean differences in frequency
between these periods for each season and in selected
regions are presented in table 1. Changes in frequency
are expressed as a percentage relative to the pre-AA
period. We also assess the choice of comparative years
by randomly selecting 100 sets of a number of years
from the pre-AA period corresponding to the length of
each AA era, then calculating the standard deviation of
the extreme-wave frequency in each set. Changes in
frequency from the pre-AA period to the AA era that
exceed one (two) standard deviation(s) are indicated
by an underscore (asterisk).
The changes in frequency are predominantly posi-
tive, indicating more frequent occurrences of highly
amplified jet-stream configurations in the AA era. Sea-
sonal and regional variations are generally consistent
with the spatial patterns of anomalies in poleward
thickness gradients shown in figures 4–7, particularly
the most robust positive trends in extreme waves over
the Atlantic and North American regions. We find a
statistically significant negative correlation (Spear-
man’s correlation = −0.30, >90% confidence)
between the seasonal, regional-mean change in thick-
ness gradient and the change in extreme-wave fre-
quency. The autumn particularly stands out in table 1,
with increases in extreme waves in all of the categories
representing the post-AA period (1995–2013 and
2000–2013), as would be expected because fall exhibits
the largest and most regionally consistent signal of AA.
The Atlantic and North American regions also stand
Figure 5. Same as figure 4but for spring (AMJ).
7
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

out, with increased frequencies in all post-AA cate-
gories. Decreased frequencies during Asian summer
are consistent with recent cooling in North-Central
Asia (figure 6), which strengthens the poleward gra-
dient, drives stronger zonal winds, and results in a
decreased |MCI|. Overall, the pattern of frequency
change is consistent with expectations of a more
amplified jet stream in response to rapid Arctic warm-
ing. Amplified jet-stream patterns are associated with
a variety of extreme weather events (i.e., persistent
heat, cold, wet, and dry) [22], thus an increase in
amplified patterns suggests that these types of extreme
events will become more frequent in the future as AA
continues to intensify in all seasons. These results may
also provide a mechanism to explain observed associa-
tions between sea-ice loss and continental heat waves
[23,29], cold spells [24,30,31], heavy snowfall [2],
and anomalous summer precipitation patterns in Eur-
ope [32].
Discussion and conclusions
The Arctic has warmed at approximately twice the rate
of the Northern mid-latitudes since the 1990s owing to
a variety of positive feedbacks that amplify green-
house-gas-induced global warming. This dispropor-
tionate temperature rise is expected to influence the
large-scale circulation, perhaps with far-reaching
effects. The North/South temperature gradient is an
important driver of the polar jet stream, thus as rapid
Arctic warming continues, one anticipated effect is a
slowing of upper-level zonal winds. It has been
Figure 6. Same as figure 4but for summer (JAS).
8
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

hypothesized that these weakened winds would cause
the path of the jet stream to become more meandering,
leading to slower Eastward progression of ridges and
troughs, which increases the likelihood of persistent
weather patterns and, consequently, extreme events
[1]. While weaker zonal winds have been observed in
response to reduced poleward temperature gradients,
the link to a wavier upper-level flow has not yet been
confirmed [31,33], although recent studies provide
strong support of a mechanism linking sea-ice loss in
the Barents/Kara Sea with amplified patterns over
Eurasia during winter [24,34] and summer [23]. We
also note that the annual-mean NOAA-tabulated
climate extreme index for the US [35] has increased by
approximately one-third in the AA era relative to pre-
AA years, though it is presently unknown whether
rapid Arctic warming is a contributing factor.
Here we provide evidence demonstrating that in
areas and seasons in which poleward gradients have
weakened in response to AA, the upper-level flow has
become more meridional, or wavier. Moreover, the
frequency of days with high-amplitude jet-stream
configurations has increased during recent years.
These high-amplitude patterns are known to produce
persistent weather patterns that can lead to extreme
weather events [22,23]. Notable examples of these
types of events include cold, snowy winters in Eastern
North America during winters of 2009/10, 2010/11,
and 2013/14; record-breaking snowfalls in Japan and
SE Alaska during winter 2011/12; and Middle-East
floods in winter 2012/2013, to name only a few.
We assess anomalies in the poleward
1000–500 hPa thickness gradient during the AA era
(since the mid-1990s) relative to climatology
(1981–2010), along with corresponding changes in the
Figure 7. Same as figure 4but for fall (OND).
9
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

zonal winds at 500 hPa and the waviness of the
500 hPa flow (|MCI|). While these time periods are
short and certainly include effects of other natural
fluctuations in the climate system, the conspicuous
emergence of AA since the mid-1990s dictates this
focused temporal analysis to identify responses of the
large-scale circulation to this ‘new’forcing. Future
work will analyze climate model projections of a future
with greater global warming and intensified AA. A
recent study [36] documents a reduction in the fre-
quency of atmospheric fronts under strong green-
house forcing, particularly in high latitudes where the
meridional temperature gradient relaxes the most,
suggestive of more persistent weather patterns.
We find that in all seasons, the regions in which the
poleward gradient weakens also exhibit weaker zonal
winds (as expected via the thermal wind relationship)
and consequently a more meridional, or wavier, flow
character. This localized response is corroborated by
seasonally varying, regional-scale increases in the fre-
quency of amplified jet-stream configurations. The
strongest response occurs during fall, when sea-ice
loss and increased atmospheric water vapor augment
Arctic warming, and a robust response is also evident
during summer over North America and the Atlantic
sectors, when the observed rapid decline of early-sum-
mer snow cover and the lower heat capacity of land
promote a drying and warming of high-latitude land
areas. Significant increases are observed in winter and
spring, as well. These results reinforce the hypothesis
that a rapidly warming Arctic promotes amplified jet-
stream trajectories, which are known to favor persis-
tent weather patterns and a higher likelihood of
extreme weather events. Based on these results, we
conclude that further strengthening and expansion of
AA in all seasons, as a result of unabated increases in
greenhouse gas emissions, will contribute to an
increasingly wavy character in the upper-level winds,
and consequently, an increase in extreme weather
events that arise from prolonged atmospheric
conditions.
Figure 8. Seasonal scatterplots relating individual gridpoint values of anomalies during 1995–2013 for zonal and meridional winds at
500 hPa (U500, V500; m s
−1
), poleward gradients in 1000–500 hPa thickness (m deg
−1
), and |MCI|(unitless). Black asterisks indicate
gridpoints corresponding to the highest 10% of total wind speed during 1995–2013 at 500 hPa, a proxy for the jet stream. Red asterisks
indicate correlations between black points that exceed 50% and p≪.001.
10
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

Acknowledgments
The authors are grateful for funding provided by the
National Science Foundation’s Arctic System Science
Program (NSF/ARCSS 1304097), to programming
assistance from R Kyle Zahn, and for helpful sugges-
tions from Dr John Walsh and two anonymous
reviewers.
References
[1] Francis J A and Vavrus S J 2012 Evidence linking Arctic
amplification to extreme weather in mid-latitudes Geophys.
Res. Lett. 39 L06801
[2] Liu J, Curry J A, Wang H, Song M and Horton R M 2012
Impact of declining Arctic sea ice on winter snowfall Proc. Natl
Acad. Sci. USA 109 4074–9
[3] Simmonds I 2015 Comparing and contrasting the behaviour
of Arctic and Antarctic sea ice over the 35 year period
1979–2013 Ann. Glaciol. 56 2015
[4] Tedesco M, Fettweis X, Mote T, Wahr J, Alexander P,
Box J E and Wouters B 2013 Evidence and analysis of 2012
Greenland records from spaceborne observations, a regional
climate model and reanalysis data Cryosphere 7615–30
[5] Derksen C and Brown R 2012 Spring snow cover extent
reductions in the 2008–2012 period exceeding climate model
projections Geophys. Res. Lett. 39 L19504
[6] Brintanja R and Selten F M 2014 Future increases in Arctic
precipitation linked to local evaporation and sea ice retreat
Nature 509 479–82
[7] Holland M M and Bitz C 2003 Polar amplification of climate
change in coupled models Clim. Dyn. 21 221–32
[8] Screen J A and Simmonds I 2010 The central role of
diminishing sea ice in recent Arctic temperature amplification
Nature 464 1334–7
[9] Alexeev V A, Langen P L and Bates J R 2005 Polar amplification
of surface warming on an aquaplanet in ‘ghost forcing’
experiments without sea ice feedbacks Clim. Dyn. 24 655–66
[10] Pithan F and Mauritsen T 2014 Arctic amplification domi-
nated by temperature feedbacks in contemporary climate
models Nat. Geosci. 7181–4
[11] Graversen R G, Langen P L and Mauritsen T 2014 Polar
amplification in CCSM4: contributions from the lapse rate and
surface Albedo feedbacks J. Clim. 27 4433–50
[12] Thomson J and Rogers W E 2014 Swell and sea in the emerging
Arctic ocean Geophys. Res. Lett. 41 3136–40
[13] Khon V C, Mokhov I I, Pogarskiy F A, Babanin A, Dethloff K,
Rinke A and Matthes H 2014 Wave heights in the 21st century
Arctic ocean simulated with a regional climate model Geophys.
Res. Lett. 41 2956–61
[14] Porter D F, Cassano J J and Serreze M C 2012 Localand large-
scale atmospheric responses to reduced Arctic sea ice and
ocean warming in the WRF model J. Geophys. Res. 117 D11115
[15] Barnes E A 2013 Revisiting the evidence linking Arctic
amplification to extreme weather in midlatitudes Geophys. Res.
Lett. 40 1–6
[16] Screen J A and Simmonds I 2013 Exploring links between
Arctic amplification and mid-latitude weather Geophys. Res.
Lett. 40 959–64
[17] Barnes E A, Dunn-Sigouin E, Masato G and Woollings T 2014
Exploring recent trends in Northern hemisphere blocking
Geophys. Res. Lett. 41 638–44
[18] Wallace J M and Hobbs P V 1977 Atmospheric Science, An
Introductory Survey (New York: Academic)
[19] Thompson D W J and Wallace J M 2001 Regional climate
impacts of the Northern hemisphere annular mode Science
293 85–9
[20] Feldstein S B and Lee S 2014 Intraseasonal and interdecadal jet
shifts in the Northern hemisphere: the role of warm pool
tropical convection and sea ice J. Clim. 27 6497–518
[21] Francis J A, Chan W, Leathers D J and Miller J R 2009 Winter
Northern hemisphere weather patterns remember summer
Arctic sea-ice extent Geophys. Res. Lett. 36 L07503
Table 1. Percentage change in seasonal frequency of high-amplitude days from the pre-AA period to the AA-era assessed using three differ-
ent initial years to define the AA-era: 1990–2013 (left column under each seasonal heading), 1995–2013 (middle column), and 2000–2013
(right column). High-amplitude days are identified when the difference between the maximum and minimum latitude of the selected daily
500 hPa height contour within a specified region exceeds 35° latitude (see text for contour selections). Underlined values indicate changes
that exceed one standard deviation of wave frequencies during the pre-AA period as determined from 100 sets of randomly selected groups
of years of the same number as the corresponding AA era (i.e., 24, 19, and 14 years beginning in 1979). Underlined and asterisked values
exceed two standard deviations.
11
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus

[22] Screen J A and Simmonds I 2014 Amplified mid-latitude
planetary waves favour particular regional weather extremes
Nat. Clim. Change 4704–9
[23] Coumou D, Petoukhov V, Rahmstorf S, Petri S and
Schellnguber H J 2014 Quasi-resonant circulation regimes and
hemispheric synchronization of extreme weather in boreal
summer Proc. Nat. Acad. Sci. 12331–6
[24] Kim B-M, Son S-W, Min S-K, Jeong J-H, Kim S-J, Zhang X,
Shim T and Yoon J-H 2014 Weakening of the stratospheric
polar vortex by Arctic sea ice loss Nat. Commun. 44646
[25] Vihma T 2014 Effects of Arctic sea ice decline on weather and
climate Surv. Geophys. 35 1175–214
[26] Kalnay E et al 1996 The NCEP/NCAR 40 year realanysis project
Bull. Am. Meteorol. Soc. 77 437–71
[27] Davini T D 2013 Atmospheric blocking and mid-latitude
climate variability PhD Dissertation, Programme in Science
and Management of Climate Change University of Foscari,
Italy
[28] Screen J A, Deser C, Simmonds I and Tomas R 2014
Atmospheric impacts of Arctic sea-ice loss, 1979–2009:
separating forced change from atmospheric internal variability
Clim. Dyn. 43 333–44
[29] Tang Q, Zhang X and Francis J A 2014 Extreme summer
weather in Northern mid-latitudes linked to a vanishing
cryosphere Nat. Clim. Change 445–50
[30] Tang Q, Zhang X, Yang X and Francis J A 2013 Cold winter
extremes in Northern continents linked to Arctic sea ice loss
Environ. Res. Lett. 8014036
[31] Cohen J et al 2014 Recent Arctic amplification and extreme
mid-latitude weather Nat. Geosci. 7627–37
[32] Screen J A 2013 Influence of Arctic sea ice on European
summer precipitation Environ. Res. Lett. 8044015
[33] Hall R, Erdelyi R, Hanna E, Jones J M and Scaife A A 2014
Review: drivers of North Atlantic polar front jet stream
variability Int. J. Climatol. doi:10.1002/joc.4121
[34] Mori M, Watanabe M, Shiogama H, InoueJ and Kimoto M
2014 Robust Arctic sea-ice influence on the frequent Eurasian
cold winters in past decades Nat. Geosci. 7869–73
[35] Karl T R, K night R W, Ea sterli ng D R and Qua yle R G 1996
Indices of climate change for the United States Bull. Am.
Meteorol. Soc. 77 279–92
[36] Catto J L, Nicholls N, Jackob C and Shelton K L 2014
Atmospheric fronts in current and future climates Geophys.
Res. Lett. 41 7642–50
12
Environ. Res. Lett. 10 (2015) 014005 J A Francis and S J Vavrus