Summer Arctic Cold Anomaly Dynamically Linked to

1

Summer Arctic Cold Anomaly Dynamically Linked to East Asian Heat Waves

2

Bingyi Wu

3 4

Department of Atmospheric and Oceanic Sciences & Institute of Atmospheric Sciences, Fudan University, Shanghai

5 6

Jennifer A. Francis

7

Woods Hole Research Center, Falmouth, Massachusetts USA

8 9

Submitted to Journal of Climate, accepted, Dec. 6, 2018

10 11 12 13 14

Corresponding author:

15

Email:

Bingyi Wu

[email protected]

16 17 18 19 20 21 22 23 1

24

Abstract:

25

During recent years, the rapidly warming Arctic and its impact on winter weather

26

and climate variability in the mid- and low-latitudes have been the focus of many

27

research efforts. In contrast, anomalous cool Arctic summers and their impacts on the

28

large-scale circulation have received little attention. In this study, we use atmospheric

29

re-analysis data to reveal a dominant pattern of summer 1000-500 hPa thickness

30

variability north of 30°N and its association with East Asian heat waves. It is found

31

that the second thickness pattern exhibits strong interannual variability but does not

32

exhibit any trend. Spatially, the positive phase of the second thickness pattern

33

corresponds with significant Arctic cold anomalies in the mid- and low-troposphere,

34

which are surrounded by warm anomalies outside the Arctic. This pattern is the

35

thermodynamic expression of the leading pattern of upper tropospheric westerly

36

variability and significantly correlated with the frequency of East Asian heat waves.

37

The Arctic has experienced frequent summer cold anomalies since 2005,

38

accompanied by strengthened tropospheric westerly winds over most of the Arctic and

39

weakened westerlies over the mid- and low-latitudes of Asia. The former significantly

40

enhances baroclinicity over the Arctic, which dynamically contributes to increased

41

frequency of anomalous low surface-pressure during summer along with decreased

42

frequency over high-latitudes of Eurasia and North America. The latter is exhibited by

43

sustained high-pressure anomalies in the mid- and low-troposphere that dynamically

44

facilitate the occurrence of East Asian heat waves. A systematic northward shift of

45

Asian zonal winds dynamically links Arctic cold anomalies with East Asian heat 2

46

waves and produces a seesaw structure in zonal wind anomalies over the Arctic and

47

the Tibetan Plateau (the third pole). Evidence suggests that enhanced Arctic westerlies

48

may provide a precursor to improve predictions of the East Asian winter monsoon,

49

though the mechanism for this lag association is unclear.

50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 3

68

1. Introduction

69

Over the past two decades, Arctic warming and Arctic sea ice loss are among the

70

most remarkable signals of change in the climate system and have been received a

71

great deal of attention, particularly in summer. The Arctic dipole anomaly, Arctic

72

anticyclonic surface wind anomalies, enhanced downwelling longwave radiation, and

73

increased Atlantic and Pacific water inflow are believed to contribute to Arctic sea ice

74

loss (Shimada et al., 2006; Wang et al., 2009; Ogi et al., 2010; Polyakov et al., 2010;

75

Overland et al., 2012; Wu et al., 2012; Ding et al., 2017). Observations and model

76

simulations demonstrate that anomalies in Arctic sea ice and atmospheric circulation

77

affect summer atmospheric circulation variability over Eurasia (Wu et al., 2009; Wu et

78

al., 2013; Screen et al., 2013). Arctic sea ice loss would reduce (enhance) summer

79

precipitation over the mid- and high-latitudes of East Asia (northern Europe) (Wu et

80

al., 2013; Screen et al., 2013). However, what are dominant features of summer Arctic

81

temperature variability in the mid- and low-troposphere that closely relates to

82

atmospheric variability in the mid- and low-latitudes of East Asia remains unclear.

83

Summer high temperature and heat waves have frequently occurred in the

84

worldwide since the beginning of 2000s, and they directly caused a high fatality and

85

produced widespread economic impacts (Meehl and Tebaldi, 2004; Barriopedro et al.,

86

2011; Bador et al., 2017). Compared with 1991-2000, casualties related with high

87

temperature and heat waves increased by more than 2000% in 2001-2010 (WMO

88

report, 2013). East Asia has experienced frequent heat waves since the 1990s (Ding et

89

al., 2010). The well documented example is the East Asian heat waves of 2013 (Min 4

90

et al., 2014; Imada et al., 2014; Zhou et al., 2014). South Korea had its hottest

91

summer nights and second hottest summer days since 1954 (Min et al., 2014), and

92

China suffered the strongest heat wave since 1951 (Zhou et al., 2014). Anthropogenic

93

global warming generally has been shown to increase the likelihood of summer heat

94

waves over East Asia (Sun et al., 2014; Min et al., 2014; Coumou et al., 2014, 2015;

95

Mann et al., 2017). Additionally, sea surface temperature (SST) anomalies, Arctic sea

96

ice loss/snow cover melting, and precipitation anomalies in India and South China

97

Sea also contribute to East Asian summer high temperature and heat waves (Hu et al.,

98

2011; Sun, 2014; Min et al., 2014; Imada et al., 2014; Tang et al., 2014; Liu et al.,

99

2015). Sun (2014) proposed that SST over the mid-North Atlantic in July 2013 was

100

the warmest in the past 160 years, which connects to weakening of the East Asian

101

upper level westerly and strengthening of the northwest Pacific subtropical high

102

(NWPSH) through a teleconnection wave train. This contributes to surface air

103

temperature variability and heat waves over the Jianghuai-Jiangnan region of China.

104

More direct causes can be attributed to sustained high pressure systems, particularly

105

the NWPSH (Wang et al., 2013; Sun, 2014; Imada et al., 2014; Wang et al., 2017;

106

Freychet et al., 2017; Gao et al., 2018). It is not clear whether simultaneous Arctic

107

atmospheric circulation anomalies during summer are linked to summer heat waves in

108

East Asia, and this is the question explored in this study. We use NCEP-NCAR

109

re-analysis data to identify dominant patterns of summer (JJA) atmospheric thickness

110

(1000-500 hPa) variability north of 30ºN and demonstrate the association between

111

summer Arctic cold anomalies and East Asian heat waves. 5

112 113

2. Data and Methods

114

Atmospheric data used in this study include monthly mean sea level pressure

115

(SLP), winds, and geopotential heights from January 1979 to December 2016, and

116

daily surface air temperatures (SATs), SLP, air temperatures and winds at 17 pressure

117

levels from 1 January 1979 to 31 December 2016. Data were obtained from the

118

National Centers for Environmental Prediction/National Center for Atmospheric

119

Research

120

(http://iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NCEP-NCAR/.CDAS-1/), with a

121

horizontal resolution of 2.5º2.5ºand 17 pressure levels (1000 to 10 hPa) in the

122

vertical direction. Daily SATs are used to calculate the frequency of summer

123

(June-July-August) extreme heat wave events, identified when daily SATs are above

124

one standard deviation for a given date and location (note similar results were

125

obtained using a 1.5 standard deviation threshold). Similarly, daily SLP data are used

126

to assess the frequency of summer anomalous low-pressure when daily SLP is below

127

-1.5 standard deviation for a given date and location.

(NCEP/NCAR)

Re-analysis

I

128

This study uses the northwestern Pacific subtropical high (NWPSH) indices,

129

which include intensity, ridge location, and western ridge point, obtained from

130

National

131

(http://cmdp.ncc-cma.net/Monitoring/cn_index_130.php). The NWPSH intensity

132

index is defined as the sum of the grid area with a geopotential height  5880 gpm

133

multiplied by the difference between the geopotential height ( 5880 gpm) minus

Climate

Center

(China)

6

134

5870 gpm within 110º-180ºE and 10º-60ºN. The NWPSH ridge location is defined

135

as the average of the latitudes at each longitude where zonal wind u =0.0 and

136

at 500 hPa within 110º-150ºE and 10º-60ºN. The NWPSH western ridge point is

137

defined as the westernmost location of the isoline with 5880 gpm within 90º-180ºE

138

and 10º-60ºN. This study also uses the monthly mean Arctic Oscillation (AO) index

139

for

140

noaa.gov/products/precip/CWlink/daily_ao_index/monthly.ao.index.b50.current.ascii)

141

.

the

period

from

1979

to

2016

𝜕𝑢 𝜕𝑦

0.0

(http://www.cpc.ncep.

142

We use summer mean 1000-500 hPa atmospheric thickness to approximately

143

represent a vertically averaged air temperature in the mid- and lower-troposphere. The

144

empirical orthogonal function (EOF) analysis method is applied to extract the first

145

two dominant patterns of summer mean 1000-500 hPa atmospheric thickness

146

variability north of 30ºN for the period 1979-2016. The first two patterns account for

147

29% and 10% of the variance. Similarly, EOF analysis is also used to analyze the

148

leading pattern of summer 300 hPa zonal wind variability north of 20ºN, which

149

accounts for 14% of the variance.

150

The maximum Eady growth rate (EGR) relates baroclinic instability to the

151

vertical wind shear (Vallis, 2006; Simmonds and Lim, 2009), characterizing the

152

conduciveness of the environment to synoptic development (Crawford and Serreze,

153

2016). EGR (𝜎𝐸 ) is given by

154 155

𝑓

𝜕𝑈

𝜎𝐸 = 0.3098 |𝑁| | 𝜕𝑧 | where f is the Coriolis parameter, U is the vertical profile of the daily zonal wind, z is 7

𝑔 𝜕𝜃

156

the vertical coordinate, and N is the Brunt-Väisäläfrequency (𝑁 2 =

157

acceleration due to gravity, θ is potential temperature). For EGR at 600 hPa, the

158

vertical profiles of θ are calculated as differences between their daily values at 500

159

and 700 hPa. The vertical shear of U is calculated in terms of the daily meridional

160

temperature gradient using the thermal wind equation.

161

3. Results

162

3.1 Atmospheric thickness variability and Arctic cold anomaly

𝜃 𝜕𝑧

, where g is

163

SATs are generally used to characterize Arctic warming and Arctic amplification.

164

Because SATs are strongly influenced by sea ice concentrations and SST, however,

165

differences in SATs between the Arctic and mid-latitudes may exaggerate the thermal

166

contrast of the column in the mid- and lower-troposphere (Sellevold et al., 2016; Wu,

167

2017). This study uses the summer 1000-500 hPa thickness to represent the mean

168

temperature in the mid- and low-troposphere (Overland and Wang, 2010; Francis and

169

Vavrus, 2015), and its first two dominant patterns are shown in Fig. 1a-c. The leading

170

pattern displays strong interannual variability superposed on an interdecadal shift, i.e.,

171

from frequent negative phases prior to 1998 to more frequent highly positive phases

172

afterward (Fig. 1a). According to the student’s t-test, the significance of the shift

173

exceeds the 99% confidence level. Spatially, positive thickness anomalies cover much

174

of the domain north of 30ºN, particularly over the Arctic, Eurasia and North America

175

where significant positive anomalies are observed (Fig. 1b). This interdecadal shift is

176

generally consistent with changes in surface wind variability over the Arctic Ocean in

177

both spring (AMJ) and summer (JAS): an anomalous cyclone prevailed prior to the 8

178

late 1990s, which was replaced by an anomalous anticyclone over the Arctic Ocean in

179

later years (Wu et al., 2012). This interdecadal shift is consistent with the rapid

180

declining trend in September sea ice extent. Thus, this leading pattern is closely

181

associated with both summer Arctic warming and sea ice loss.

182

Here we focus primarily on the second thickness pattern because it is associated

183

with summer high temperatures and heat waves in Asia and North America. This

184

pattern displays strong interannual variability but does not exhibit any trend or shift

185

(Fig. 1a). Spatially (Fig. 1c), negative thickness anomalies are observed in the Arctic

186

north of 70ºN, indicating Arctic cold anomalies in the mid- and low-troposphere.

187

Positive thickness anomalies appear mainly over northern Europe, northeastern

188

Atlantic, western North America, the Bering Sea, north-central Canada, and mid- and

189

low-latitudes of East Asia. Over the eastern Pacific-North America and Northern

190

Pacific-East Asia, positive anomalies form a “comma” structure, which is associated

191

with the upper tropospheric steering flow anomalies (see the following section). All

192

extreme positive anomalies occurred after 2005, i.e., 2006, 2013, and 2016, with

193

values exceeding 1.5 σ, corresponding with Arctic cold anomalies (Figs 1d-f). The

194

negative phases of PC2 dominated during 2007-2012, implying summer Arctic warm

195

anomalies (Fig. 1c), consistent with the well-documented rapid Arctic sea ice loss

196

period. Years exhibiting summer Arctic cold anomalies seem to be become more

197

frequent in the context of Arctic warming. It should be pointed out that two thickness

198

patterns are independent from each other and the leading thickness pattern cannot

199

obscure the expression of the second thickness pattern. Additionally, although summer 9

200

500 hPa height anomalies exhibit a great similarity to 1000-500 hPa thickness

201

anomalies (Figs. 1c, 2), differences are also visible, including extents and amplitudes

202

of positive and negative anomalies.

203

3.2 Dynamic linkages between Arctic cold anomalies and East Asian heat waves

204

Significant correlations between the leading thickness pattern and frequency of

205

summer heat waves are mainly confined to near the Ural Mountains (40-80°E and

206

40-60°N), North America, and the Arctic, rather than over East Asia (not shown).

207

Cold anomalies are also observed over near Alaska, but they are not significant (Fig.

208

1b). This study, therefore, focuses on the second thickness pattern. The positive phase

209

of PC2 correlates significantly with heat waves over the mid- and low-latitudes of

210

Asia, particularly over the Tibetan Plateau (the third pole) and the coast of East Asian

211

(Fig. 3a). Significant correlations also exist over most of Europe, northern Canada,

212

southwestern United States, and northern North Pacific Ocean. Decreased heat waves

213

are observed over the Arctic Ocean, consistent with Arctic cold anomalies (Fig. 1c).

214

Thus, a dipole structure is apparent between the Arctic and several mid-latitude

215

continental areas, particularly with the third pole. Over East Asia, the regionally

216

(80.625 º -135 º E, 25.71 º -39.05 º N) averaged frequency of summer heat waves

217

exhibits an increasing trend at the 99% confidence level (Fig. 3b). The frequency of

218

East Asian heat waves exceeded 1.0 σ in 1994, 2001, 2006, 2010, 2013, and 2016

219

(Fig. 3b). Consequently, East Asian heat waves frequently occurred after 2005. It

10

220

should be pointed out that if the criterion for heat waves is instead defined as  1.5

221

standard deviations, very similar results are obtained (not shown).

222

Why do Arctic cold anomalies correspond with more intense and frequent

223

summer heat waves in the mid- and low-latitudes of East Asia? We find that this

224

association is closely related to the large-scale upper tropospheric zonal wind

225

variability. The positive phase of the second thickness pattern significantly

226

strengthens upper tropospheric westerlies over most of the Arctic (Fig. 4a), while over

227

Eurasia, wind anomalies display a belt structure with significantly weaker westerlies

228

in the mid- and low-latitudes of Asia and Europe. This wind pattern closely resembles

229

the leading pattern of summer 300 hPa zonal wind variability north of 20ºN (not

230

shown, r = 0.81; Fig. 4b). The area-weighted, regionally averaged 300 hPa westerly

231

north of 70ºN (Arctic westerly index) is also significantly correlated with the second

232

thickness pattern (r = 0.88, Fig. 4c). In the upper troposphere, Arctic westerly strength

233

is out of phase with that in the mid- and low-latitudes of Asia. The normalized Arctic

234

westerly indices exceeded 1.5 σ in 2006, 2013, and 2016, corresponding with strong

235

East Asian heat waves (Fig. 3b). Over the mid- and low-latitudes of Asia, weakened

236

upper tropospheric steering flow (Fig. 4d) favors the stagnation of air-masses and

237

dynamically contributes to sustained high pressure anomalies in the mid- and

238

low-troposphere

239

lower-troposphere, thereby reducing cloud cover and enhancing the downwelling

240

surface shortwave radiation, which favors high temperatures and heat waves. Sun

241

(2014) also showed a significantly negative correlation (r = -0.72) between July SAT

(Fig.

4e).

These

anomalies

suppress

convection

in

the

11

242

averaged over the region bounded by 110º-125ºE and 27.5º-32.5ºN and the 200 hPa

243

zonal wind averaged over the same region, consistent with our results here.

244

Positive Arctic westerly anomalies are not confined to the upper troposphere;

245

they are evident throughout the troposphere and stratosphere (Fig. 5). Negative

246

westerly anomalies in the mid- and low-latitudes also penetrate through the

247

troposphere and stratosphere. Such a change is attributed to a systematic northward

248

shift of zonal winds, dominantly characterized by a shift of the East Asian westerly jet.

249

In fact, the Eurasian westerly jet systematically migrates northward (Fig. 4a).

250

Additionally, a dipole structure in zonal wind anomalies is also observed over the

251

Arctic and the third pole (Figs 4a, 4d, 5), similar to the pattern in temperature

252

anomalies discussed previously. Over the Arctic, negative temperature anomalies are

253

completely confined to the below 300 hPa and top level of the stratosphere, with

254

positive temperature anomalies between them (Fig. 6). Tropospheric temperature

255

anomalies in the mid- and low-latitudes are in phase with Arctic temperature

256

anomalies in the upper troposphere and much of stratosphere. Strengthening of

257

tropospheric westerly winds may enhance baroclinicity in the mid- and

258

low-troposphere,

259

low-pressure during summer. Figure 7 supports this deduction. We find that

260

significant increases in baroclinicity are observed over the Arctic Ocean, surrounded

261

by negative anomalies outside the Arctic (Fig. 7a). Decreases in baroclinicity emerge

262

over high-latitudes of Eurasia and North America, northern North Pacific, and East

263

Asia, generally consistent with warm areas shown in Figs 1c, 3a. Significant increases

which

dynamically

favors

the

occurrence

of

anomalous

12

264

in the frequency of anomalous low-pressure are confined to the Arctic Ocean and

265

northern North Atlantic, while negative anomalies cover most of Eurasia and North

266

America (Fig. 7b). Over high-latitudes of the continents, decreases in the frequency

267

of anomalous low-pressure may imply that thermal exchanges between the Arctic

268

Ocean and lower-latitudes weaken. It should be noted that if the threshold used to

269

define anomalous low-pressure is set to below -1.0 or -2.0 standard deviations in daily

270

SLP anomalies, very similar results are obtained (not shown).

271

4. Discussion

272

4.1 Roles of NWPSH in East Asian heat waves

273

Although the NWPSH is believed to contribute to East Asian heat waves (Luo

274

and Lau, 2017; Gao et al., 2018), some major characteristics of this subtropical high,

275

including intensity and location, are not significantly correlated with the second

276

thickness pattern or with the Arctic westerly index. We find that correlation

277

coefficients of the second thickness pattern with the NWPSH intensity, ridge location,

278

and western ridge point indices are 0.07, -0.09, and -0.24, respectively. While East

279

Asia experienced abnormal high temperature and heat waves in 2013, the NWPSH

280

was in a neutral phase (Fig. 8). The statistical association between the NWPSH and

281

heat waves also differs from that in Fig. 3a (not shown). Thus, increasingly frequent

282

East Asian heat waves cannot be simply attributed to the NWPSH.

283

4.2 Possible dynamic processes linking Arctic cold anomalies 13

284

We assess the statistical relationship between summer Arctic cold anomalies in

285

the mid- and low-troposphere and East Asian heat waves. As previously noted, we

286

find strengthened westerly winds in the Arctic along with weakened zonal winds in

287

the mid- and low-latitudes of East Asia, which we attribute to a systematic northward

288

shift of the zonal wind belts. The present study, however, cannot identify the

289

mechanism responsible for this shift nor for the temporal/spatial behavior of the

290

second thickness pattern. It is possible that natural variability may be a major reason

291

for recent changes in this pattern.

292

Here we carry out the case analysis of summer 2013 (Fig. 1) to explore possible

293

reasons for summer Arctic cold anomalies in terms of wave-flow interactions at 500

294

hPa:

295

̅ 𝜕𝜃 𝜕𝑡

̅ 𝜕𝜃

= −𝑉̅ ∙ ∇𝜃̅ − 𝜔 ̅ 𝜕𝑝 − ̅̅̅̅̅̅̅̅̅̅ ∇ ∙ 𝜃′𝑉′ −

̅̅̅̅̅̅̅ ′ 𝜔′ 𝜕𝜃 𝜕𝑝

+ 𝐹̅

(1)

296

where  is potential temperature, V is horizontal wind vector, and 𝐹̅ is the forcing

297

term (Hoskins and Pearce, 1983). The overbar and prime denote time mean and

298

transient eddy terms, respectively. For this case, the time mean is defined as the mean

299

over 2013 JJA (92 days), and the transient eddy is obtained by subtracting the time

300

mean for a given date and location. Summer 500 hPa temperature anomalies are

301

shown in Fig. 9; the spatial pattern closely resembles the thickness anomaly pattern

302

shown in Fig. 1e. Negative anomalies are confined to the Arctic, surrounded by

303

positive anomalies to the south. Summer 500 hPa potential temperature anomalies

304

closely resemble that in Fig. 9 (not shown). We then calculated the contributions of 14

̅̅̅̅̅̅̅ ′ 𝜔′

305

𝜕𝜃 ̅ ̅̅̅̅̅̅̅̅̅̅ each term (−𝑉̅ ∙ ∇𝜃̅, −𝜔 ̅ 𝜕𝜃 , −∇ ∙ 𝑉 ′𝜃′, − 𝜕𝑝

306

tendencies (Fig. 10). We find that both positive and negative contributions from each

307

term occur simultaneously in the Arctic, and negative contributions mean that both

308

mean potential temperature advection by the mean flow as well as transient eddy flux

309

divergence contribute to summer Arctic cold anomalies. Compared to the transient

310

eddy flux divergence, the vertical advection of mean potential temperature plays a

311

more important role in driving Arctic cold anomalies. Over the mid- and low-latitudes

312

of Asia and northwestern Pacific, negative vertical advections are predominant except

313

for central and eastern China where the vertical advection is positive, contributing to

314

increases in mean potential temperature (Figs 6, 9). The summer mean vertical

315

velocities at 500 hPa during 2013 (Fig. 11) suggest that vertical motions dominate the

316

contribution of vertical advection of mean temperature (Fig. 10b). We repeat the same

317

analysis process for 2006 and 2016 summers, producing similar results (not shown).

318

This implies that Arctic upward motion in the mid- and low-troposphere is a major

319

reason for summer Arctic cold anomalies. Ogi et al. (2004) investigated summer

320

annual mode and its association with anomalies in zonal mean temperature and

321

meridional circulation and found that summer Arctic cold anomalies in the mid- and

322

low-troposphere correspond to upward motion anomalies (see their Fig. 3d),

323

consistent with results here. Although anomalous atmospheric circulation patterns

324

discussed by Ogi et al. (2004) exhibit some similarities to the second thickness pattern,

325

their differences are robust (Fig. 12). We also note that downward and upward motion

326

co-exist simultaneously over the Arctic (Fig. 11) and their roles are opposite.

𝜕𝑝

) to the mean potential temperature

15

327

4.3 Precursor to predict East Asian winter monsoon

328

Finally, we find that the strengthened summer Arctic westerly is significantly

329

correlated with the ensuing Asian winter climate variability (Fig. 13). The correlation

330

between the Arctic westerly index and the winter Siberian high index is -0.59,

331

significant at 99% confidence level (Fig. 13a). At 500 hPa, geopotential height

332

anomalies exhibit a tripole structure, and the positive height anomalies cover East

333

Asia, indicating a weakened East trough (Fig. 13b). This configuration resembles the

334

so-called Eurasian pattern (Wallace and Gutsler, 1981; Wang and Zhang, 2015).

335

Negative SLP anomalies occupy the mid- and high-latitudes of Eurasia (Fig. 13c).

336

Thus, winter atmospheric circulation anomalies associated with strengthened Arctic

337

westerly winds result in anomalous winter warming and a weakened winter monsoon

338

over East Asia (Fig. 13d). Thus, the summer Arctic westerly index is an important

339

precursor for winter surface air temperature anomalies over East Asia and it may

340

prove to be useful predictor. It is not clear, however, which mechanisms are

341

responsible for this lagged relationship. It is possible that summer high temperatures

342

and strong heat waves in central and eastern Asia produce anomalies in soil

343

temperature and moisture that lead to a weakened East Asian winter monsoon. This

344

question requires further investigation.

345

5. Conclusions

346

This study reveals the first two dominant patterns of summer 1000-500 hPa

347

thickness variability north of 30°N, and they respectively account for 29% and 10% 16

348

of the variance. The leading thickness pattern displays strong interannual variability

349

superposed on an interdecadal shift that occurred in the late 1990s, consistent with

350

summer Arctic warming and rapid Arctic sea ice loss. Spatially, positive thickness

351

anomalies cover much domain north of 30ºN, particularly over the Arctic, Eurasia

352

and North America where significant positive anomalies are observed. The second

353

thickness pattern exhibits strong interannual variability but does not exhibit any trend

354

or shift. The positive phase of the second thickness pattern corresponds with

355

significant Arctic cold anomalies in the mid- and low-troposphere, which are

356

surrounded by warm anomalies outside the Arctic. Arctic cold anomalies have

357

occurred more frequently in the context of Arctic warming (2005-2016).

358

The second thickness pattern is the thermodynamic expression of the leading

359

pattern of upper tropospheric westerly variability and shows significant positive

360

correlations with frequencies of East Asian heat waves. In the upper troposphere,

361

enhanced Arctic westerly winds are out of phase with winds in the mid- and

362

low-latitudes of Asia. The stronger Arctic westerly winds significantly enhance

363

baroclinicity, which dynamically contributes to increased frequency of anomalous

364

low-pressure over the Arctic. The weaker westerly winds favor stagnation of warm air

365

and dynamically contributes to sustained high pressure anomalies in the mid- and

366

low-troposphere, increasing the likelihood of East Asian heat waves. These wind

367

anomalies can be attributed to a systematic northward shift of zonal winds,

368

dominantly characterized by a shift of the East Asian westerly jet. This shift

369

dynamically produces a dipole structure in zonal wind anomalies over the Arctic and 17

370

the third pole. Dynamic analysis indicates that Arctic upward motion in the mid- and

371

low-troposphere is a major reason for summer Arctic cold anomalies. The enhanced

372

Arctic westerly and Arctic cold anomalies during summer may provide a precursor to

373

predict East Asian winter monsoon.

374

Acknowledgements

375

We thank Prof. Jianqi Sun, and two anonymous reviewers for their insightful

376

comments, which significantly improved this manuscript. The authors are grateful to

377

NCEP/NCEP for providing atmospheric re-analysis data. BW was supported by the

378

National Natural Science Foundation of China (41730959, 41790472, and 41475080)

379

and the National Key Basic Research Project of China (2015CB453200), and JF was

380

supported by NASA grant NNX14AH896 and funding from the Woods Hole Research

381

Center.

382 383

384

References

385

Bador, M., L. Terray, J. Boe, S. Somot, A. Alias, A. Gibelin, and B. Dubuisson, 2017:

386

Future summer mega-heatwave and record-breaking temperatures in a warmer

387

France climate. Environmental Research Letters. 12(7), 074025.

388

Barriopedro, D., E. Fischer, J. Luterbacher, R. Trigo, and R. García-Herrera, 2011:

389

The hot summer of 2010: Redrawing the temperature record map of Europe.

390

Science, 332, 220-224.

18

391

Coumou, D., V. Petoukhov, S. Rahmstorf, S. Petri, and H. Schellnhuber, 2014:

392

Quasi-resonant circulation regimes and hemispheric synchronization of ex-

393

treme weather in boreal summer. Proc. Natl. Acad. Sci. USA, 111, 12331–

394

12336.

395

Coumou, D., J. Lehmann, and J. Beckmann, 2015: The weakening summer circulation

396

in the Northern Hemisphere mid-latitudes. Science, 348(6232), 324-327.

397 398 399

Crawford, A.D., and M.C. Serreze, 2016: Does the summer arctic frontal zone influence arctic ocean cyclone activity? J. Climate, 29, 4977-4993. Ding, Q., and coauthors, 2017: Influence of high-latitude atmospheric circulation

400

changes on summertime Arctic sea ice. Nature Climate Change, 7, 289-295.

401

Ding, T., W. Qian, and Z. Yan, 2010: Changes in hot days and heat waves in China

402 403 404 405

during 1961-2007. International Journal of Climatology, 30(10), 1452-1462. Francis, J., and S. Vavrus, 2015: Evidence for a wavier jet stream in response to rapid Arctic warming. Environ. Res. Lett., 10, 014005. Freychet, N., S. Tett, J. Wang, and G. Hegerl, 2017: Summer heat waves over Eastern

406

China: Dynamical processes and trend attribution. Environmental Research

407

Letters, 12(2), 024015.

408

Gao, M., B. Wang, J. Yang, and W. Dong, 2018: Are peak summer sultry heat wave

409

days over the Yangtze-Huaihe river basin predictable? J. Clim., 31,

410

2185-2196.

411 412

Hoskins, B.J., and R.P. Pearce, 1983: Large-scale Dynamical Processes in the Atmosphere. Academic Press (London), PP397. 19

413

Hu, K., G. Huang, and R. Huang, 2011: The impact of tropical Indian Ocean

414

variability on summer surface air temperature in China. J. Clim., 24(20),

415

5365-5377.

416

Imada, Y., H. Shiogama, M. Watanabe, M. Mori, M. Ishii, and M. Kimoto, 2014: The

417

contribution of anthropogenic forcing to the Japaness heat waves of 2013. Bull.

418

Amer. Meteor. Soc., S52-54.

419

Liu, G., R. Wu, S. Sun, and H. Wang, 2015: Synergistic contribution of precipitation

420

anomalies over northwestern India and the South China Sea to high

421

temperature over the Yangtze River valley. Advances in Atmospheric Sciences,

422

32, 1255-1265.

423 424

425

Luo, M., and N. Lau, 2017: Heat waves in southern China: Synoptic behavior, long-term change and urbanization effects. J. Clim., 30(2), 703–720. Mann, M., S. Rahmstorf, K. Kornhuber, B. Steinman, S. Miller, and D. Coumou,

426

2017: Influence of anthropogenic climate change on planetary wave resonance

427

and extreme weather events. Sci. Rep., 7, 45242.

428 429

Meehl, G., and C. Tebaldi, 2004: More intense, more frequent, and longer lasting heat waves in the 21st century. Science. 305, 994-997.

430

Min, S.-K., Y.-H. Kim, M.-K. Kim, and C. Park, 2014: Assessing human contribution

431

to the summer 2013 Korea heat wave. Bull. Amer. Meteor. Soc., S48-51.

432

Ogi, M., K. Yamazaki, and Y. Tachibana, 2004: The summertime annular mode in the

433

Northern Hemisphere and its linkage to the winter mode, J. Geophys. Res.,

434

109, D20114, doi:10.1029/2004JD004514. 20

435

Ogi, M., K. Yamazaki, and J. Wallace, 2010: Influence of winter and summer surface

436

wind anomalies on summer Arctic sea ice extent. Geophys. Res. Lett., 37,

437

L07701. DOI: 10.1029/ 2009GL042356.

438 439

Overland, J., and M. Wang, 2010: Large-scale atmospheric circulation changes are associated with the recent loss of Arctic sea ice. Tellus, 62A, 1-9.

440

Overland, J., J. Francis, H. Edward, and M. Wang, 2012: The recent shift in early

441

summer Arctic atmospheric circulation. Geophys. Res. Lett., 39, L19804,

442

doi:10.1029/2012GL053268.

443 444 445 446 447 448

Polyakov, I., and co-authors, 2010: Arctic Ocean warming contributes to reduced polar ice cap. J. Phys. Oceanogr., 40, 2743-2756. Screen, J., 2013: Influence of Arctic sea ice on European summer precipitation. Environ. Res. Lett., 8, doi: 10.1088/1748-9326/8/4/044015. Sellevold, R., S. Sobolowski, and C. Li, 2016: Investigating possible Arctic– midlatitude teleconnections in a linear framework. J. Clim., 29, 7329–7343.

449

Shimada, K., T. Kamoshida, M. Itoh, S. Nishino, E. Carmack, and co-authors, 2006:

450

Pacific Ocean inflow: influence on catastrophic reduction of sea ice cover in

451

the

452

10.1029/2005GL025624.

Arctic

Ocean.

Geophys.

Res.

Lett.,

33,

L08605.

DOI:

453

Simmonds, I., and E.P. Lim, 2009: Biases in the calculation of South Hemisphere

454

mean baroclinic eddy growth rate. Geophys. Res. Lett., 36, L01707,

455

doi:10.1029/2008GL036320.

456

Sun, J. Q., 2014: Record-breaking SST over mid-North Atlantic and extreme high 21

457

temperature over the Jianghuai-Jiangnan region of China in 2013. Chin. Sci.

458

Bull., 59(27), 3465-3470, doi:10.1007/s11434-014-0425-0.

459 460 461 462

Sun, Y. et al., 2014: Rapid increase in the risk of extreme summer heat in Eastern China. Nat. Clim. Change, 4(12), 1082-1085. Tang, Q., X. Zhang, and J. Francis, 2014: Extreme summer weather in northern mid-latitudes linked to a vanishing cryosphere. Nat. Clim. Change, 4, 45-50.

463

Vallis, G.K., 2006: Atmospheric and Oceanic Dynamics: Fundamentals and

464

Large-Scale Circulation. Cambridge Univ. Press, New York.

465

Wallace, J.M., and D.S. Gutzler, 1981: Teleconnections in geopotential height field

466

during the Northern Hemisphere winter. Monthly Weather Review, 109,

467

784-812.

468

Wang, J., J. Zhang, E. Watanabe, M. Ikeda, K. Mizobata, and co-authors, 2009: Is the

469

dipole anomaly a major driver to record lows in Arctic summer sea ice extent?

470

Geophys. Res. Lett., 36, L05706. DOI: 10.1029/2008GL036706.

471

Wang, N., and Y. Zhang, 2015: Evolution of Eurasian teleconnection pattern and its

472

relationship to climate anomalies in China. Climate Dynamics, 44, 1017-1028.

473

Wang, P., J. Tang, X. Sun, S. Wang, J. Wu, X. Dong, and J. Fang, 2017: Heat waves

474

in China: definitions, leading patterns, and connections to large-scale

475

atmospheric circulation and SSTs. J. Geophys. Res. Atmos., 122,

476

10679-10699.

477 478

Wang, W., W. Zhou, X. Wang, S. Fong, and K. Leong, 2013: Summer high temperature extremes in Southeast China associated with the East Asian jet 22

479

stream and circumglobal teleconnection. J. Geophys. Res. Atmos., 118,

480

8306-8319.

481 482 483

WMO, 2013: The global climate 2001-2010, a decade of climate extremes. WMO-No.1103, 119pp. Wu, B., R. Zhang, B. Wang, and R. D’Arrigo, 2009: On the association between

484

spring Arctic sea ice concentration and Chinese summer rainfall. Geophys. Res.

485

Lett., 36, L09501, doi:10.1029/2009GL037299.

486

Wu, B., J. Overland, and R. D’Arrigo, 2012: Anomalous arctic surface wind patterns

487

and their impacts on September sea ice minima and trend. Tellus, 64A, 18590,

488

doi:10.3402/tellusa.v64i0.18590.

489

Wu, B., R. Zhang, R. D’Arrigo, and J. Su, 2013: On the relationship between winter

490

sea ice and summer atmospheric circulation over Eurasia. Journal of Climate,

491

26, 5523-5536.

492 493 494

Wu, B., 2017: Winter atmospheric circulation anomaly associated with recent Arctic winter warm anomalies. J. Clim., 30, 8469-8479. Zhou, T., S. Ma, and L. Zou, 2014: Understanding a hot summer in central Eastern

495

China: summer 2013 in contest of multimodel trend analysis. Bull. Amer.

496

Meteor. Soc., S54-57.

497

498

499

Figure captions

23

500

Figure 1 The first two patterns of summer (June-July-August, JJA) 1000-500 hPa

501

thickness variability north of 30ºN. (a) Normalized PC time series (red: PC1; blue:

502

PC2), the red dashed line represents a linear trend in PC1, and the blue dashed line

503

indicates the mean of PC2 averaged over 2007-2012. (b) Summer 1000-500 hPa

504

thickness anomalies, derived from a linear regression on the detrended PC1, the

505

white and black contours represent thickness anomalies at 95% and 99%

506

confidence levels, respectively. (c) As in (b), but for the regression on the

507

normalized PC2. (d)-(f) summer 1000-500 hPa thickness anomalies (relative to the

508

mean averaged over 1979-2016) in 2006 (d), 2013 (e), and 2016 (f). The first two

509

patterns respectively account for 29% and 10% of the variance.

510

Figure 2 summer mean 500 hPa geopotential height anomalies, derived from a linear

511

regression on the normalized PC2. White and black contours represent anomalies at

512

95% and 99% confidence levels, respectively.

513

Figure 3 (a) Regression map of the frequency of summer heat wave events, regressed

514

on the normalized PC2 of atmospheric thickness variability. White and black

515

contours denote 95% and 99% confidence levels, respectively. (b) Regionally

516

averaged heat wave frequencies (normalized time series) over the domain bounded

517

by the green box in (a). The dashed line represents a linear trend at 99% confidence

518

level in (b).

519

Figure 4 Weakened tropospheric westerly steering flow links East Asian heat waves

520

to Arctic cold anomalies. (a) Spatial distribution of summer 300 hPa zonal wind

521

anomalies, derived from a linear regression on the normalized PC2. White and 24

522

black contours represent anomalies at 95% and 99% confidence levels, respectively;

523

the green line represents 0º-180ºE at 30ºN. (b) Normalized PC2 (blue line) and

524

PC1 of summer 300 hPa zonal wind variability north of 20ºN (red line; the leading

525

wind pattern accounts for 14% of the variance; r=0.81 ). (c) As in (b), but for

526

normalized, area-weighted, and regionally averaged 300 hPa zonal winds north of

527

70ºN (red line). Correlation between the two time series is 0.88. (d) As in (a), but

528

for zonal wind anomalies along the longitude-pressure cross section at 30ºN (green

529

line in (a)). The gray area indicates topography along 30ºN. (e) As in (d), but for

530

geopotential height anomalies.

531

Figure 5 A systematic shift northward of zonal winds in the troposphere and

532

stratosphere. Zonal wind anomalies along the latitude–pressure cross section at 110

533

ºE, derived from a linear regression on the normalized PC2. White and black

534

contours represent anomalies at 95% and 99% confidence levels, respectively.

535

Green contours denote zonal winds averaged over 1979-2016, and purple contours

536

are zonal winds averaged over Arctic cold summers, with the normalized PC2  1,

537

i.e., 1989, 1994, 1996, 2006, 2013, and 2016.

538

Figure 6 Air temperature anomalies along the latitude–pressure cross section at 110º

539

E, derived from a linear regression on the normalized PC2. White and black

540

contours represent anomalies at 95% and 99% confidence levels, respectively.

541

Figure 7 (a) Spatial distribution of summer mean EGR anomalies at 600 hPa (day-1),

542

derived from a linear regression on the normalized Arctic westerly index. White

543

and black contours denote anomalies at 95% and 99% confidence levels, 25

544

respectively. (b) As in (a), but for anomalies in the frequency of summer anomalous

545

low-pressure.

546

Figure 8 (a) Normalized time series of the NWPSH intensity index, (b)-(c), As in (a),

547

but for the ridge location (latitude) and western ridge point (longitude), respectively.

548

All

549

(http://cmdp.ncc-cma.net/Monitoring/cn_index_130.php).

550 551

552

data

obtained

from

National

Climate

Center

in

Figure 9 2013 summer mean 500 hPa temperature anomalies (relative to the summer mean averaged over 1979-2016) (ºC). Figure 10 500 hPa mean temperature advection and time mean of transient eddy flux ̅

553

𝜕𝜃 ̅̅̅̅̅̅̅̅̅̅ divergence for summer of 2013: (a) −𝑉̅ ∙ ∇𝜃̅, (b) −𝜔 ̅ 𝜕𝑝, (c) −∇ ∙ 𝜃 ′ 𝑉 ′ , (d)

554

−

̅̅̅̅̅̅̅ ′ 𝜔′ 𝜕𝜃 𝜕𝑝

, and (e) sum of (a)-(d). Units are K day-1.

555

Figure 11 Mean vertical velocity at 500 hPa (Pa s-1) for summer of 2013.

556

Figure 12 Normalized time series: summer mean AO index (red) and second

557

China

thickness pattern (blue), and their correlation is 0.54.

558

Figure 13 (a) Normalized time series of the summer Arctic westerly index (red line)

559

and the ensuing winter (DJF) Siberian high index (blue line). Their correlation is

560

-0.59, significant at 99% confidence level. (b) Winter 500 hPa geopotential height

561

anomalies (gpm), derived from a linear regression on the normalized summer

562

Arctic westerly index. White and black contours represent anomalies at 95% and 99%

563

confidence levels, respectively. (c)-(d), As in (b), but for winter SLP (hPa) and SAT

564

(ºC) anomalies, respectively.

26

565

566

Figures

567

568

27

569 570

Figure 1 The first two patterns of summer (June-July-August, JJA) 1000-500 hPa

571

thickness variability north of 30ºN. (a) Normalized PC time series (red: PC1; blue:

572

PC2), the red dashed line represents a linear trend in PC1, and the blue dashed line

573

indicates the mean of PC2 averaged over 2007-2012. (b) Summer 1000-500 hPa

574

thickness anomalies, derived from a linear regression on the normalized detrended

575

PC1, the white and black contours represent thickness anomalies at 95% and 99%

576

confidence levels, respectively. (c) As in (b), but for the regression on the

577

normalized PC2. (d)-(f) summer 1000-500 hPa thickness anomalies (relative to the 28

578

mean averaged over 1979-2016) in 2006 (d), 2013 (e), and 2016 (f). The first two

579

patterns respectively account for 29% and 10% of the variance.

580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604

29

605 606

Figure 2 summer mean 500 hPa geopotential height anomalies, derived from a linear

607

regression on the normalized PC2. White and black contours represent anomalies at

608

95% and 99% confidence levels, respectively.

609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 30

628

629 630

631 632

Figure 3 (a) Regression map of the frequency of summer heat wave events, regressed

633

on the normalized PC2 of atmospheric thickness variability. White and black

634

contours denote 95% and 99% confidence levels, respectively. (b) Regionally

635

averaged heat wave frequencies (normalized time series) over the domain bounded

636

by the green box in (a). The dashed line represents a linear trend at 99% confidence

637

level in (b). 31

638 639

640

641

642 32

643

644

645 646 647 648 33

649

Figure 4 Weakened tropospheric westerly steering flow links East Asian heat waves

650

to Arctic cold anomalies. (a) Spatial distribution of summer 300 hPa zonal wind

651

anomalies, derived from a linear regression on the normalized PC2. White and

652

black contours represent anomalies at 95% and 99% confidence levels, respectively;

653

the green line represents 0º-180ºE at 30ºN. (b) Normalized PC2 (blue line) and

654

PC1 of summer 300 hPa zonal wind variability north of 20ºN (red line; the leading

655

wind pattern accounts for 14% of the variance; r=0.81 ). (c) As in (b), but for

656

normalized, area-weighted, and regionally averaged 300 hPa zonal winds north of

657

70ºN (red line). Correlation between the two time series is 0.88. (d) As in (a), but

658

for zonal wind anomalies along the longitude-pressure cross section at 30ºN (green

659

line in (a)). The gray area indicates topography along 30ºN. (e) As in (d), but for

660

geopotential height anomalies.

661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 34

681 682 683 684 685 686

687 688 689

Figure 5 A systematic shift northward of zonal winds in the troposphere and

690

stratosphere. Zonal wind anomalies along the latitude–pressure cross section at 110

691

ºE, derived from a linear regression on the normalized PC2. White and black

692

contours represent anomalies at 95% and 99% confidence levels, respectively.

693

Green contours denote zonal winds averaged over 1979-2016, and purple contours

694

are zonal winds averaged over Arctic cold summers, with the normalized PC2  1,

695

i.e., 1989, 1994, 1996, 2006, 2013, and 2016.

696

35

697

698

699 700

Figure 6 Air temperature anomalies along the latitude–pressure cross section at 110º

701

E, derived from a linear regression on the normalized PC2. White and black

702

contours represent anomalies at 95% and 99% confidence levels, respectively.

703 704 705

36

706 707 708

Figure 7 (a) Spatial distribution of summer mean EGR anomalies at 600 hPa (day-1),

709

derived from a linear regression on the normalized Arctic westerly index. White

710

and black contours denote anomalies at 95% and 99% confidence levels,

711

respectively. (b) As in (a), but for anomalies in the frequency of the anomalous

712

low-pressure.

713

37

714 715

Figure 8 (a) Normalized time series of the NWPSH intensity index, (b)-(c), As in (a),

716

but for the ridge location (latitude) and western ridge point (longitude), respectively.

717

All

718

(http://cmdp.ncc-cma.net/Monitoring/cn_index_130.php).

data

obtained

from

National

Climate

Center

in

China

719

720

721

722

38

723

724 725

Figure 9 2013 summer mean 500 hPa temperature anomalies (relative to the summer mean averaged over 1979-2016) (ºC).

39

726 727

Figure 10 500 hPa mean potential temperature advection and time mean of transient

728

𝜕𝜃 ̅̅̅̅̅̅̅̅̅̅ eddy flux divergence for summer of 2013: (a) −𝑉̅ ∙ ∇𝜃̅, (b) −𝜔 ̅ 𝜕𝑝, (c) −∇ ∙ 𝜃′𝑉 ′,

729

(d) −

̅

̅̅̅̅̅̅̅ ′ 𝜔′ 𝜕𝜃 𝜕𝑝

, and (e) sum of (a)-(d). Units are K day-1.

730 40

731

732

733

Figure 11 Mean vertical velocity at 500 hPa (Pa s-1) for summer of 2013.

734

735

736

737

738

739

740 41

741

742

743

744

745

746

747 748

Figure 12 Normalized time series: summer mean AO index (red) and second thickness pattern (blue), and their correlation is 0.54.

749

750

751

752

753

754

755

42

756

757

758

759

760

43

761

762 763

Figure 13 (a) Normalized time series of the summer Arctic westerly index (red line)

764

and the ensuing winter (DJF) Siberian high index (blue line). Their correlation is

765

-0.59, significant at 99% confidence level. (b) Winter 500 hPa geopotential height

766

anomalies (gpm), derived from a linear regression on the normalized summer 44

767

Arctic westerly index. White and black contours represent anomalies at 95% and 99%

768

confidence levels, respectively. (c)-(d), As in (b), but for winter SLP (hPa) and SAT

769

(ºC) anomalies, respectively.

770

45