The impact of atmospheric Rossby waves and cyclones on …

Climate Dynamics manuscript No.(will be inserted by the editor)

The impact of atmospheric Rossby waves and cyclones on theArctic sea ice variability

Marte G. Hofsteenge · Rune G. Graversen · Johanne H. Rydsaa · Zoe

Rey

Received: date / Accepted: date

Abstract The Arctic sea-ice extent has strongly de-1

clined over recent decades. A large inter-annual vari-2

ability is superimposed on this negative trend. Previous3

studies have emphasised a significant warming effect as-4

sociated with latent energy transport into the Arctic re-5

gion, in particular due to an enhanced greenhouse effect6

associated with the convergence of the humidity trans-7

port over the Arctic. The atmospheric energy transport8

into the Arctic is mostly accomplished by waves such9

as Rossby waves and cyclones. Here we present a sys-10

tematic study of the effect on Arctic sea ice of these at-11

mospheric wave types. Through a regression analysis we12

investigate the coupling between transport anomalies of13

both latent and dry-static energy and sea-ice anomalies.14

From the state-of-the-art ERA5 reanalysis product the15

latent and dry-static transport over the Arctic bound-16

ary (70 ◦N) is calculated. The transport is then split17

M. G. HofsteengeSchool of Geography, University of Otago, Dunedin, NewZealandDepartment of Physics and Technology, UiT The Arctic Uni-versity of Norway, Tromsø, Norway, Orcid ID: 0000-0001-5369-5318 E-mail: [email protected]

Rune G. GraversenDepartment of Physics and Technology, UiT The Arctic Uni-versity of Norway, Tromsø, NorwayNorwegian Meteorological Institute,Tromsø, Norway

Johanne H. RydsaaDepartment of Physics and Technology, UiT The Arctic Uni-versity of Norway, Tromsø, Norway, Orcid ID: 0000-0002-0538-0714

Zoe ReyDepartment of Physics and Technology, UiT The Arctic Uni-versity of Norway, Tromsø, Norway

into transport by planetary and synoptic-scale waves 18

using a Fourier decomposition. The results show that 19

latent energy transport as compared to that of dry- 20

static shows a much stronger potential to decrease sea 21

ice concentration. However, taking into account that 22

the variability of dry-static transport is of an order of 23

magnitude larger than latent, the actual impact on the 24

sea ice appears similar for the two components. In ad- 25

dition, the energy transport by planetary waves causes 26

a strong decline of the sea ice concentration whereas 27

the transport by synoptic-scale waves shows only little 28

effect on the sea ice. The study emphasises the impor- 29

tance of the large-scale waves on the sea ice variability. 30

Keywords Arctic sea ice, atmospheric energy 31

transport, Rossby waves, cyclones, wave decomposition, 32

Arctic climate 33

1 Introduction 34

The fast decrease in Arctic sea ice extent is one of the 35

clearest indicators of the ongoing global warming due to 36

anthropogenic greenhouse gas emissions (Fox-Kemper 37

et al., 2021). Passive microwave satellite records from 38

1978 and onwards show significant trends of a decreas- 39

ing sea-ice extent for every month of the year, with 40

September showing the strongest decrease (Serreze and 41

Stroeve, 2015). Superimposed on this decreasing trend 42

is a large inter-annual variability. As of today, the record- 43

low sea-ice extent during the observational period was 44

reached at the end of summer of 2012, although the 45

2020 September extent of only 3.92 million square kilo- 46

meters was 2.49 million square kilometers below the 47

1981 to 2010 average, and was close to break the 2012 48

record. 49

2 Marte G. Hofsteenge et al.

Several processes and mechanisms have been sug-50

gested to be important for the sea-ice variability, such51

as those induced by radiative effects related to sur-52

face albedo (Kashiwase et al., 2017), low-level clouds53

(Kay et al., 2008; Kay and L’Ecuyer, 2013; Liu and54

Schweiger, 2017), surface winds (Ogi et al., 2010; Zhang55

et al., 2013; Mills and Walsh, 2014), and oceanic (Arthun56

et al., 2012) and atmospheric (Kapsch et al., 2013; Park57

et al., 2015; Woods and Caballero, 2016; Wang et al.,58

2020) energy transport. Olonscheck et al. (2019) stud-59

ied the contribution of different processes on the sea-ice60

variability based on experiments with the ECHAM661

climate model, eliminating processes one-by-one, and62

demonstrated that the sea-ice variability is mainly con-63

trolled by atmospheric temperature fluctuations induced64

by atmospheric energy transport or local ocean heat65

release, whereas variability of surface albedo, clouds,66

water vapour, surface winds, and ocean heat transport67

play a minor role. Previous studies have shown impact68

of atmospheric heat and moisture transport variability69

on the Arctic surface temperatures (Graversen, 2006;70

Woods et al., 2013; Baggett et al., 2016), and that71

e.g. cloud, humidity, and surface albedo are also modi-72

fied by this transport (Graversen and Burtu, 2016; Liu73

and Schweiger, 2017; Graversen and Langen, 2019).74

In addition, several studies have documented link-75

age between the mid-latitude atmospheric circulation76

and individual sea-ice events. A large sea-ice anomaly77

during summer of 2007 was encountered in the vicin-78

ity of the East-Siberian Sea, and was associated with79

anomalous atmospheric energy transport through the80

Pacific sector (Graversen et al., 2011). The warm and81

humid air that was advected into the Arctic resulted82

in anomalous longwave radiation and turbulent fluxes83

towards the surface in the area where the ice melt was84

encountered. Also the cloud-cover over the ice-melt area85

was affected (Schweiger et al., 2008). Another extreme86

year when it comes to Arctic summer sea-ice melt is,87

as mentioned above, 2012. It has been argued that this88

event was at least partly caused by an intense storm89

over the Arctic in the beginning of August that year90

(Simmonds and Rudeva, 2012; Zhang et al., 2013). Also91

during the winter season warm and humid advection92

from lower latitude appear to have a negative impact93

on sea-ice extent. March 2017 showed one of the low-94

est Arctic ice-extent levels for that month of the year95

since 1979, which is argued to be linked to episodes of96

increased warm and humid atmospheric transport from97

lower latitudes during the 2016-2017 autumn and win-98

ter season, which hampered the ice growth (Hegyi and99

Taylor, 2018).100

101

Analysis including several years of data show that 102

atmospheric circulation in spring is important for the 103

development of the sea-ice over the summer. For years 104

with a low September sea ice extent, an anomalous con- 105

vergence of moisture transport over the Arctic in spring 106

induces an enhanced greenhouse effect, which causes 107

strong melt early in the melt season, which through 108

the ice-albedo feedback can accelerate the melt over the 109

summer (Kapsch et al., 2013, 2016; Yang and Magnus- 110

dottir, 2017). Also it has been suggested that precondi- 111

tioning associated with unusual atmospheric circulation 112

hampering ice growth in the cold season can effect sea- 113

ice melt in the succeeding summer season, since under 114

these circumstances the ice is already thinner at the 115

beginning of the melt season (Doscher and Koenigk, 116

2013). Also variation in the onset of the Arctic melt 117

season appears to be dependent on transport of mois- 118

ture into the Arctic (Mortin et al., 2016), which affect 119

the local cloudiness (Liu and Schweiger, 2017). 120

121

It has also been argued that short-lived synoptic- 122

scale storm events in winter can cause large sea-ice re- 123

treats. The moisture that is brought into the Arctic 124

by these storms strongly dampens the radiative cooling 125

and the ice growth (Graham et al., 2019). In addition, 126

ice growth is hampered due to snowfall from the storm 127

events, since snow acts as an effective insulator, and 128

the associated strong winds fracture ice and enhances 129

ocean atmosphere heat exchange and lateral melt. Also 130

Boisvert et al. (2016) noted that humid and warm air 131

transported by cyclonic activity in winter of 2015-2016 132

led to decreased sea-ice conditions that winter. How- 133

ever, Screen et al. (2011) found that low cyclone activ- 134

ity in late spring and summer is linked to a reduced 135

September sea-ice cover. 136

137

Hence, although it appears well documented that 138

atmospheric circulation and its northward energy trans- 139

port play an important role for the Arctic sea-ice vari- 140

ability, there is little consensus as to the role of the 141

underlying circulation processes. Here we investigate 142

the individual effect of large-scale planetary waves and 143

synoptic-scale cyclones. Planetary waves have the po- 144

tential to impact large part of the Arctic, whereas cy- 145

clones induce more local effects by nature. Earlier stud- 146

ies have indicated that planetary waves impact Pan- 147

Arctic temperatures more than do synoptic waves (Baggett148

and Lee, 2015; Graversen and Burtu, 2016), which, as 149

will be shown here, also is the case for the impact on 150

sea ice. 151

152

From the state-of-the-art ERA5 reanalysis product 153

we calculate atmospheric energy transport split into a 154

The impact of atmospheric Rossby waves and cyclones on the Arctic sea ice variability 3

dry-static and a latent energy component, where the155

latter is associated with the transport of water vapour.156

The latent component as compared to the dry-static157

has been shown to have a stronger impact on the Arctic158

surface-air temperatures per unit energy (Woods et al.,159

2013; Graversen and Langen, 2019). The transport can160

be further decomposed into planetary and synoptic-161

scale waves based on a Fourier decomposition (Gra-162

versen and Burtu, 2016). Planetary waves as compared163

to synoptic scale waves appear to have a stronger im-164

pact on Arctic climate (Baggett and Lee, 2015; Gra-165

versen and Burtu, 2016), although there is a strong in-166

terplay between the two components (Baggett et al.,167

2016; Papritz and Dunn-Sigouin, 2020). We investigate168

here the relationship between the different transport169

components and sea ice anomalies using lagged regres-170

sions and composites of extreme transport events.171

2 Methods and data172

The study is focused on the transport of dry-static and173

latent energy over the Arctic boundary, which is defined174

here as the 70◦ N latitude. We use the state-of-the-art175

ERA5 (Hersbach et al., 2020) reanalysis products with176

a 0.5 degrees horizontal spatial resolution, 137 verti-177

cal hybrid levels, and 6-hourly time resolution for the178

period 1979-2018. Due to a mass-inconsistency of re-179

analysis data (Trenberth, 1991), a barotropic mass-flux180

correction has been applied to the wind fields at each 6-181

hourly time prior to the calculation of the energy trans-182

ports (Graversen, 2006).183

184

2.1 Decomposition of the atmospheric energy185

transport186

Based on this reanalysis product, both the zonally and187

vertically integrated dry-static and latent energy trans-188

port are calculated as,189

vD(φ) =

∮ ∫ ps

0

v

(cpT + gz +

1

2v2

)dp

gdx (1)

and

vQ(φ) =

∮ ∫ ps

0

vLqdp

gdx, (2)

respectively, where φ is latitude, v=(u, v) the zonal190

and meridional wind components, cp the specific heat191

capacity at constant pressure, T temperature, g grav-192

ity, p pressure, ps surface pressure, L the latent heat of193

condensation and q specific humidity. 194

195

The transport can be further split into transport 196

by planetary and synoptic-scale waves using a Fourier 197

decomposition (Graversen and Burtu, 2016; Heiskanen 198

et al., 2020). The decomposition is performed by apply- 199

ing a Fourier transformation of v dpg , (cpT +gz + v2/2), 200

and Lq before taking the zonal and vertical integrals. 201

For Lq the transformation gives: 202

Lq =aL02

+

∞∑n=1

{aLn cos

(n2πx

d

)+ bLn sin

(n2πx

d

)},

(3)

where d = 2πR cos(φ), R is the Earth’s radius, φ lati- 203

tude, n the zonal wave number. The Fourier coefficients 204

aLn and bLn are given as: 205

aLn =2

d

∮Lq cos

(n2πx

d

)dx (4)

and

bLn =2

d

∮Lq sin

(n2πx

d

)dx, (5)

respectively, where aLn and bLn depend on latitude, height 206

and time. Applying the Fourier transformation in a sim- 207

ilar way to v dpg and (cpT + gz + v2/2) allows for a 208

decomposition of the transport into parts for different 209

wave numbers. As suggested by Heiskanen et al. (2020), 210

planetary waves are defined as wave numbers 1-3 and 211

synoptic-scale waves numbers 4-20, as it was shown that 212

this separation was most similar to a split between the 213

two types of waves at 70 ◦N applying an independent 214

method based on a wavelet decomposition. This Fourier 215

decomposition applied here is complete in the sense that 216

the sum of the transport of waves 1-20 and the merid- 217

ional transport, wave 0, gives a negligible residual rela- 218

tive to the total transport (Graversen and Burtu, 2016). 219

220

A decomposition of the vertical integrated trans- 221

port as a function of longitude in addition to latitude 222

and time can be obtained by applying the Fourier de- 223

composition only to the mass transport. For the latent 224

transport of planetary and synoptic waves, this can be 225

expressed: 226

227

vQP (λ, φ) =

Nv∑k=1

{3∑

n=1

[avn cos

(n2πx

d

)+ bvn sin

(n2πx

d

)]Lq

},


(6)

and228

vQS(λ, φ) =

Nv∑k=1

{∑n≥4

[avn cos

(n2πx

d

)+ bvn sin

(n2πx

d

)]Lq

},

(7)

respectively, where λ is longitude, and avn and bvn are229

the Fourier coefficients for the mass flux, v dpg , computed230

similar to Eqs. 4 and 5, respectively, and Nv = 137 is231

the number of vertical levels in ERA5.232

2.2 Regressions233

In order to study the impact of the different transport234

components on the Arctic sea-ice extent, we perform235

a regression analysis of the sea-ice concentration (SIC)236

on the transport components. The ERA5 SIC data are237

based on the HadISST2.0.0.0 and OSI SAF SIC prod-238

ucts (Hirahara et al., 2016). We use daily SIC data with239

a 1-degree spatial resolution. The annual cycle and the240

trend in the SIC data were removed by subtracting a241

5-year running mean climatology over the period from242

1979 til 2018. The running mean was computed similar243

to Kapsch et al. (2019), where the first and last two244

years of the period are calculated as a weighted mean.245

By subtracting the running mean climatology this way,246

the long-term trend is removed even if not linear. The247

SIC anomalies including year-to-year variability are re-248

tained.249

250

The seasonal cycle was removed from the trans-251

port data, to isolate the impact of transport anomalies.252

Transport data is not de-trended and a test with de-253

trended data showed that this would not influence the254

regression coefficients we found. The energy transport255

influences sea ice through their effect on the surface256

energy components. Therefore surface energy fields in-257

cluding shortwave radiative, longwave radiative, sensi-258

ble and latent heat fluxes are used from the ERA5 prod-259

uct, where positive values indicate downward fluxes. For260

the surface energy fields, trend and seasonal cycle are261

removed by subtracting a 5-year running-mean clima-262

tology, similarly as for the SIC data.263

264

All regressions are calculated using the 40-years time-265

series (1979-2018). The regressions are computed for266

different positive and negative time-lags, to find the267

statistical link of events occurring before and after each268

other. The regression coefficients are calculated as the 269

Beta-coefficient of linear regression. To calculate the re- 270

gression (R) of y on x, we use: 271

272

R =

∑(x− x)(y − y)∑

(x− x)2(8)

The significance of the regression coefficients are 273

tested using a Monte Carlo approach. In this approach 274

5000 random time series were computed by applying 275

a random phase shift to the Fast Fourier Transform of 276

the time-series. This way of randomizing the data series 277

makes sure that the artificial time-series has the same 278

power spectrum as the original time-series. The regres- 279

sion coefficients that are calculated using the artificial 280

time-series are then compared with the coefficients for 281

the original data. We use a 95 and 99 % significance 282

level, meaning that when less than 5 and 1 %, respec- 283

tively, of the artificial regression coefficients are numer- 284

ically larger than the original regressions, the regression 285

coefficient is considered significant. 286

3 Results 287

3.1 Impact of planetary and synoptic transport on 288

Arctic sea ice 289

Atmospheric energy-transport by planetary waves leads 290

to a decrease in Arctic sea-ice concentration (SIC). This 291

is evident from the regressions of SIC on the dry-static 292

and latent transport component (Fig. 1a and b). For 293

positive time lag, transport by these wave types show 294

significant negative regressions, meaning that planetary 295

transport events are followed by a decrease in SIC. 296

297

The regression coefficients for the latent transport 298

are clearly larger compared to those of the dry-static. 299

This indicates that for anomalous transport events of 300

similar magnitude, the latent transport has a consider- 301

ably larger impact on the Arctic sea ice than its dry- 302

static counterpart. This is in agreement with previ- 303

ous finding that the latent transport as compared to 304

the dry-static affects the Arctic surface-air temperature 305

more (Graversen and Burtu, 2016). The regression co- 306

efficients indicate that a latent transport anomaly by 307

planetary waves of 1 PW results in a decrease in sea 308

ice area of about 0.22 million km2 around 4 days after 309

the anomalous event. For an average Arctic sea-ice area 310

of about 10 million km2 over the 40-year study period, 311

this is a decrease of 2.2 % of the ice area. The signifi- 312

cant negative anomalies sustain for about 9 days after 313

a dry-static transport event, but up to 30 days after 314


−40 −30 −20 −10 0 10 20 30 40

−0.20

−0.15

−0.10

−0.05

0.00

0.05

0.10

0.15

Regressio

n coefficient [m

illion km

2 /PW]

a) Dry-static

Significant (95%)Significant (99%)Pla e%arySy op%ic

−40 −30 −20 −10 0 10 20 30 40

−0.20

−0.15

−0.10

−0.05

0.00

0.05

0.10

0.15

b) La%e %

Sig ifica % (95%)Sig ifica % (99%)Pla e%arySy op%ic

−40 −30 −20 −10 0 10 20 30 40Time lag [days]

−0.03

−0.02

−0.01

0.00

0.01

0.02

Regressio

coefficie % [m

illio km

2 ]

c) Dry-static scaled

Significant (95%)Significant (99%)Pla e%arySy op%ic

−40 −30 −20 −10 0 10 20 30 40Time lag [days]

−0.03

−0.02

−0.01

0.00

0.01

0.02d) La%e % scaled

Sig ifica % (95%)Sig ifica % (99%)Pla e%arySy op%ic

Fig. 1 Regressions of Arctic SIC on (a, c) dry-static and (b, d) latent atmospheric energy transport across 70 ◦N, as afunction of time lag. Regressions are given for both planetary and synoptic-scale waves, and are in units of Arctic SIC anomalyin million km2 per PW of transport. Frame c and d show the regression coefficients as in a and b but multiplied by thestandard deviation of the transport. Hence these frames show the sea-ice change induced by a 1 standard deviation anomaly ofthe transport components across 70 ◦N. Regressions significant on the 95 and 99 % level are shown with light and dark greenshading, respectively.

a latent event of the same magnitude. This is consis-315

tent with long-term impact that sea-ice anomalies may316

have through dynamic and radiative effects such as the317

ice-albedo feedback mechanism (Kashiwase et al., 2017;318

Graham et al., 2019).319

320

Although the latent energy transport by planetary321

waves shows a considerably stronger potential to de-322

crease Arctic sea ice than its dry-static counterpart, the323

two components appear to have similar sea-ice impact324

when the difference in variability is taken into account325

(Fig. 1c and d). The daily dry-static and latent trans-326

port by planetary waves across 70 ◦N constitute in an327

annual mean about 1.5 and 0.3 PW, respectively, and328

their standard deviations are 0.94 and 0.13 PW, re-329

spectively. By scaling the regression coefficients by the330

standard deviation of the transport components, the331

actual impact on the sea ice of the different transport332

components can be compared.333

−30 −20 −10 0 10 20 30Time lag [days]

−0.5

−0.4

−0.3

−0.2

−0.1

0.0

0.1

0.2

Regr

essio

n co

e ic

ient

[milli

on km

2 /PW

]

Non-signi icant (99%)MAMJJASONDJF

Fig. 2 As Fig. 1 but for seasonal regressions of Arctic SICon the latent energy transport by planetary waves across 70◦N as a function of time lag. Regressions significant at the 99% level are shown in color, while non-significant regressionsare indicated in black.


−40 −30 −20 −10 0 10 20 30 40

−5

0

5

10

15

Regression coefficient [W 2/PW]

a) Dry-staticSWSD, planetarySWSD, synopticLWSD, planetaryLWSD, synopticnon-significant (99%)

(40 (30 (20 (10 0 10 20 30 40

(5

0

5

10

15

b) LatentSWSD, planetarySWSD, %ynopticLWSD, planetaryLWSD, %ynopticnon-significant (99%)

−40 −30 −20 −10 0 10 20 30 40Time lag [days]

−1.5

−1.0

−0.5

0.0

0.5

1.0

1.5

2.0

Regression coefficient [W 2]

c) Dry-static scaledSWSD, planetarySWSD, synopticLWSD, planetaryLWSD, synopticnon-significant (99%)

−40 −30 −20 −10 0 10 20 30 40Time lag [days]

−1.5

−1.0

−0.5

0.0

0.5

1.0

1.5

2.0d) Latent scaled

SWSD, planetarySWSD, synopticLWSD, planetaryLWSD, synopticnon-significant (99%)

Fig. 3 As Fig. 1, but for the regressions over the Arctic sea-ice (> 15 % concentration) of shortwave (SWSD) and longwave(LWSD) downward radiation at the surface on the energy-transport components at 70 ◦N. Units are radiation anomalies inW/m2 per PW of transport. Regressions significant at the 99 % level are indicated in colour, and non-significant regressionsin black.

The impact on Arctic SIC by synoptic-scale waves334

is considerably smaller and, in fact, opposite in sign335

as compared to that of the planetary waves (Fig. 1).336

Around lag zero, significant positive correlation coeffi-337

cients are apparent for both the dry-static and latent338

synoptic-scale transport. The positive sea-ice anoma-339

lies around lag zero indicate cold Arctic conditions and340

hence an anomalously large meridional temperature gra-341

dient between mid and high latitudes which enhances342

baroclinicity in the Arctic boundary regions. These anoma-343

lous temperature contrasts strengthen the baroclinicity344

and growth of synoptic-scale waves at the Arctic bound-345

ary, which is likely leading to the positive anomalies of346

synoptic-scale waves. Hence the interpretation of cause347

and effect based on these lag-regression analyses be-348

comes opposite for planetary and synoptic-scale waves:349

Where the lag-regressions reveal that SIC reduction fol-350

lows energy transport by planetary waves, the positive351

zero-lag regressions for synoptic-scale waves are consis-352

tent with a cold Arctic leading to increase of synoptic-353

scale wave at the Arctic boundary due to enhanced 354

baroclinicity. This interpretation is consistent with pre- 355

vious findings based on temperature regressions on the 356

energy transport components (Graversen and Burtu, 357

2016). 358

359

The decrease in SIC after transport events by the 360

planetary-wave latent energy transport is the strongest 361

in the winter season (DJF; Fig. 2). The negative SIC 362

impact is significant in all seasons except summer. The 363

ice-edge region is the largest in winter and smallest in 364

summer, and therefore the transport anomalies can im- 365

pact sea ice in a larger area during the cold seasons than 366

during summer. Note that all months show positive re- 367

gressions indicating enhanced sea-ice extent prior to the 368

transport event (at lag zero), although only in spring 369

(MAM) these are significant. The increased sea-ice con- 370

centrations preceding the planetary transport events in- 371

dicate that transport extremes tend to occur predomi- 372


−40 −30 −20 −10 0 10 20 30 40−4

−2

0

2

4

6

8

Regressio

n coeffic

ient [W

m2/PW

]

a) Dry-staticSWS, planetarySWS, synopticLWS+LH+SH, planetaryLWS+LH+SH, synopticNon-significant (99%)

−40 −30 −20 −10 0 10 20 30 40−4

−2

0

2

4

6

8b) Latent

SWS, planetarySWS, synopticLWS+LH+SH, planetaryLWS+LH+SH, synopticNon-significant (99%)

−40 −30 −20 −10 0 10 20 30 40Time lag [ a−s]

−0.50

.0.25

0.00

0.25

0.50

0.75

1.00

1.25

Regressio

n co

effic

ient [W

m2]

c) Dr−-static scaledSWS, planetarySWS, synopticLWS+LH+SH, planetaryLWS+LH+SH, synopticNon-significant (99%)

−40 −30 −20 −10 0 10 20 30 40Time lag [ a−s]

−0.50

.0.25

0.00

0.25

0.50

0.75

1.00

1.25

d) Latent scale SWS, planetarySWS, synopticLWS+LH+SH, planetaryLWS+LH+SH, synopticNon-significant (99%)

Fig. 4 As Fig. 1, but for the regressions over Arctic sea ice of net shortwave (SWS) and longwave (LWS) radiation plusturbulent fluxes of sensible (SH) and latent (LH) downward at the surface on the energy-transport components at 70 ◦N.Units are radiation anomalies in W/m2 per PW of transport. Regressions significant at the 99 % level are indicated in colour,and non-significant regressions in black.

nantly during cold Arctic conditions.373

374

Both dry-static and latent energy transport across375

70 ◦N lead to convergence of heat in the atmosphere376

over the Arctic, which is partly radiated and turbu-377

lently mixed to the surface. Hereby anomalous advec-378

tion into the Arctic of the two transport components379

can induce anomalous energy flux to the surface, which380

may melt or hampers freeze up of sea ice. However,381

events of anomalous transport by the latent compo-382

nent have a much stronger potential to impact the sea383

ice because of the enhanced humidity and cloud forma-384

tion that this transport component gives rise to (Gra-385

versen and Burtu, 2016), which considerably increases386

the radiation to the surface by the greenhouse effect.387

This is evident from regressions of downwelling long-388

wave radiation (LWSD) on the two components of the389

planetary transport (Fig. 3a and b). These regressions390

show a significant increase of the LWSD after the trans-391

port events, which is considerably larger and last longer392

for the latent than for the dry-static component. Simi- 393

lar regressions for the downwelling shortwave radiation 394

(SWSD) indicate negative values for the planetary la- 395

tent transport at positive time lags consistent with an 396

increase of the cloud cover succeeding events of this 397

transport. After scaling the regressions with the stan- 398

dard deviation of the two transport components, to take 399

into account that variability of the dry-static transport 400

is an order of magnitude larger than that of the latent, 401

the actual impact on the downwelling longwave radi- 402

ation of the two transport components appear similar 403

(Fig. 3c and d). 404

405

Similar to the regression of the planetary-scale trans- 406

port, regressions of LWSD on the synoptic-scale trans- 407

ports show positive values for positive time lag, al- 408

though these regressions are smaller than those of the 409

planetary-scale transport and not significant. However, 410

significant but negative regressions are found for the 411

synoptic transport dry-static component around lag zero. 412


−40 −30 −20 −10 0 10 20 30 40−0.3

−0.2

−0.1

0.0

0.1

0.2

0.3

0.4

0.5

Regres

sion co

effic

ient [m

s−1 /P

W]

a) Dry-staticSignificant (95%)Significant (99%)P aneta%(S(noptic

)40 )30 )20 )10 0 10 20 30 40)0.3

)0.2

)0.1

0.0

0.1

0.2

0.3

0.4

0.5b) Latent

Significant (95%)Significant (99%)P aneta%(S(noptic

)40 )30 )20 )10 0 10 20 30 40Time ag [da(s]

)0.04

)0.02

0.00

0.02

0.04

0.06

0.08

Reg%

essio

n co

effic

ient

[ms)

1 ]

c) D%(-static scaledSignificant (95%)Significant (99%)PlanetarySynoptic

−40 −30 −20 −10 0 10 20 30 40Time ag [days]

−0.04

−0.02

0.00

0.02

0.04

0.06

0.08d) Latent sca ed

Significant (95%)Significant (99%)PlanetarySynoptic

Fig. 5 As Fig. 1 but for the regressions over the Arctic sea-ice (> 15 % concentration) of 10 m wind speed on the energy-transport components at 70 ◦N. Units are wind anomalies in m/s per PW of transport. Regressions significant at the 99 %level are indicated in colour, and non-significant regressions in black.

This is again consistent with the discussion above that413

enhanced cyclogenesis leading to a positive anomaly414

of the synoptic-scale waves, is associated with anoma-415

lously cold Arctic conditions with a cold atmosphere416

radiating less than usual to the surface.417

418

The increase in downwelling longwave radiation at419

the surface following planetary events may lead to a420

surplus of energy at the surface causing ice melt or re-421

duced ice growth. This is supported from regressions of422

the total surface net longwave radiation plus turbulent423

fluxes on the transport components (Fig. 4). The sum424

of net longwave radiation and turbulent fluxes consti-425

tutes the major part of the surface energy fluxes. A426

surplus of energy is found following planetary trans-427

port events, which is consistent with the ice reduction428

during the same period (Fig. 1). The net shortwave ra-429

diation is the remaining part that in conjunction with430

the net longwave radiation plus turbulent fluxes consti-431

tute the total energy transfer to the surface from the432

atmosphere. However this component is little affected433

by the transport anomalies. 434

435

Consistent with the discussion above, scaling of the 436

transport components with their variability indicates 437

that dry-static and latent planetary transport have a 438

similar impact on the surface-energy balance over the 439

Arctic, and that synoptic-scale wave events appear as- 440

sociated with an anomalously negative surface-energy 441

balance and positive sea-ice anomalies (Fig. 1). 442

443

In addition to thermodynamic effects, as shown by 444

Fig. 4, the mass flow associated with the transports 445

may also directly impact sea ice due to friction. Increase 446

in 10 m winds over the sea ice is succeeding events of 447

planetary-wave transport of both dry-static and latent 448

energy (Fig. 5). The impact per PW of transport is 449

much stronger for the latent than for the dry-static 450

component (Fig. 5a and b), since a given mass flow 451

potentially can transport far more energy of the dry- 452

static than of the latent type, and hence the mass flow 453

has to be considerably stronger to transport the same 454


amount of energy in the latent than in the dry-static455

form. Again, when the two transport components are456

scaled by the standard deviation, the wind impact be-457

comes similar (Fig. 5c and d). Note that the duration458

of the dynamical impact of the planetary-scale waves is459

shorter than that related to thermodynamic effect asso-460

ciated with anomalous surface-energy imbalance. Hence461

the response of the Arctic atmosphere to a transport462

anomaly seems to stay for another 1-2 weeks after the463

flow anomaly has relaxed.464

465

For the synoptic-wave transport, the wind anoma-466

lies associated with this transport at around lag zero467

are negative though not significant for the latent part.468

This is indicating that the atmosphere over the Arctic469

sea ice is anomalously little affected by advection and470

hence left to adiabatic cooling consistent sea-ice grow471

and in line with the discussion above.472

3.2 Regions of high sea-ice variability473

We continue by focusing on two regions that show large474

sea-ice variability, both on the annual and year-to-year475

time scales: the Chukchi and East-Siberian Sea and the476

Barents and Kara Sea regions (Kapsch et al., 2013;477

Parkinson and Cavalieri, 2008; Peng and Meier, 2018).478

The former region contributes 22 % of the summer sea-479

ice variability of the Arctic (Onarheim et al., 2018), and480

is also the region of the largest ice retreat during the481

2007 September sea-ice minimum; an event caused by482

an anomalous flow of warm and humid air into that re-483

gion (Graversen et al., 2011). The latter region accounts484

for the largest fraction of ice variability in March (27485

%, Onarheim et al. (2018)). We will here focus on the486

latent energy transport, as our previous results showed487

that this type of transport has a much larger potential488

than the dry-static to decrease the Arctic SIC.489

490

The ice concentration in the Barents and Kara Seas491

is impacted significantly by the planetary wave latent492

transport. This is evident in Fig. 6a, showing negative493

regressions for positive time lags, with a maximum im-494

pact on the sea ice around 4 days after a transport495

event. Hence positive anomalies of the planetary waves496

cause significant ice reduction that last for more than497

50 days after an event. The transport impact on sea ice498

is found in all season except summer when little ice is499

present in this region (Fig. 7). In the remaining seasons,500

the Barents and Kara Seas are part of the marginal ice501

zone, and the region is situated just north of the main502

gateway for transport events, which will be further dis-503

cussed below. It has earlier been reported that Bar-504

ents and Kara Seas have the largest SIC variability and505

−40 −30 −20 −10 0 10 20 30 40−0.125

−0.100

−0.075

−0.050

−0.025

0.000

0.025

0.050

Regr

essio

n co

effic

ient

[milli

on km

2 /PW

]

a)PlanetaryBarents/Kara SeaEast-Siberian SeaSignificant (95%)Significant (99%)

−40 −30 −20 −10 0 10 20 30 40T me lag [da)s]

−0.125

−0.100

−0.075

−0.050

−0.025

0.000

0.025

0.050

Regr

ess o

n co

eff c

en(

[m ll

on km

2 /PW

]

b)S)no%( c

Baren(s/Kara SeaEas(-Siberian SeaSignificant (95%)Significant (99%)

Fig. 6 Regressions of SIC in the Barents and Kara Seas(red) and Chukchi and East-Siberian Seas (blue) on the la-tent energy transport by (a) planetary and (b) synoptic-scalewaves across 70 ◦N. Regressions are given as a function oftime lag in units of SIC anomaly in million km2 per PW oftransport. Regressions significant on a 95 and a 99 % levelare shown with light and dark-green shading, respectively.

trends in winter (Onarheim et al., 2018). The signifi- 506

cant regressions for negative time lags in spring show a 507

pre-conditioning effect: planetary transport events are 508

more likely to occur in spring when initially there is 509

more sea ice in the Barents and Kara Seas, and hence 510

likely colder condition in this area and stronger tem- 511

perature gradient towards the Atlantic sector. 512

513

In contrast to the Barents and Kara Seas, we find 514

no significant impact on sea ice in the Chukchi and 515

East-Siberian Seas (Fig. 6a). This weak impact may ap- 516

pear surprising given that this is a region where sum- 517

mer ice minima often occur. The Chukchi and East- 518

Siberian Seas have played a major role in recent record 519

years of low summer sea ice and for the overall neg- 520

ative sea-ice trend during this season (Kapsch et al., 521

2013). However, for example when it comes to the large 522

sea-ice anomaly of 2007, which mainly appeared in this 523

area, it was found that although warm and humid air 524

from the Pacific played a major role, the zonal-mean 525

transport was not particularly strong during this sea- 526


−30 −20 −10 0 10 20 30

−0.25

.0.20

.0.15

.0.10

.0.05

0.00

0.05

0.10

0.15

Regres

sion o

effi

ient [m

illion

km

2 /PW]

a) MAMBarents/Kara SeaEast-Siberian SeaSignificant (95%)Significant (99%)

−30 −20 −10 0 10 20 30

−0.25

.0.20

.0.15

.0.10

.0.05

0.00

0.05

0.10

0.15b) JJA

Barents/Kara SeaEast-Siberian SeaSignificant (95%)Significant (99%)

−30 −20 −10 0 10 20 30T%me lag [da−s]

−0.25

.0.20

.0.15

.0.10

.0.05

0.00

0.05

0.10

0.15

Regres

sion o

effi

ient [m

illion

km

2 /PW]

c) SONBarents/Kara SeaEast-Siberian SeaSignificant (95%)Significant (99%)

−30 −20 −10 0 10 20 30T%me lag [da−s]

−0.25

.0.20

.0.15

.0.10

.0.05

0.00

0.05

0.10

0.15d) DJF


Fig. 7 As Fig. 6a but for a split into seasons, (a) March-May, (b) June-August, (c) September-November, and (d) December-February.

ice minimum event (Graversen et al., 2011). Little or527

no impact is found from the synoptic wave transport in528

neither of the investigated regions (Fig. 6b).529

3.3 Atlantic and Pacific energy transport530

The analysis so far is based on regressions of SIC on the531

zonal-mean transport across 70 ◦N. But as appeared532

from the summer-of-2007 anomaly (Graversen et al.,533

2011), the impact of atmospheric transport extremes534

on SIC might be depending on the longitude of the535

transport, and not on the zonal-averaged quantity. For536

this reason the transport is now split into Atlantic and537

Pacific-sector parts and the focus will be only on the538

latent transport by planetary waves as this component539

shows the largest sea-ice impact. The Atlantic and Pa-540

cific sectors are chosen following Rydsaa et al. (2021)541

finding that these sectors are dominating the transport542

when it comes to extreme events in the winter season.543

These sectors encounter longitudes from 30 ◦W to 30 544

◦E and from 145 ◦W to 160 ◦E, respectively, and are 545

also known for being dominant pathways for winter cy- 546

clones (Serreze et al., 1993; Zhang et al., 2004). 547

548

Indeed, when analyzing the transport over the At- 549

lantic and Pacific sector separately, both the Barents 550

and Kara Seas as well as the Chukchi and East-Siberian 551

Sea region show a sea-ice decline after the transport 552

events in the adjacent sector (Fig. 8). The ice reduc- 553

tion is consistent with enhanced thermodynamic sur- 554

face forcing associated with a net surface imbalance of 555

net longwave radiation and turbulent fluxes in favour of 556

the surface, and to some extent dynamic forcing from 557

increase in surface winds (Fig. 9). Moreover it appears 558

that a transport event in one sector, having a nega- 559

tive impact on sea ice in the adjacent region, has a re- 560

verse impact on the other, remote region. This indicates 561

a tendency of cross Arctic transport associated with 562


−40 −30 −20 −10 0 10 20 30 40−0.125

−0.100

−0.075

−0.050

−0.025

0.000

0.025

0.050

Regressio

n coefficient [m

illion km

2 /PW]

a)Atlantic


−40 −30 −20 −10 0 10 20 30 40Time la [da)s]

−0.125

−0.100

−0.075

−0.050

−0.025

0.000

0.025

0.050

Re ressi%

n c%efficien( [m

illi%n km

2 /PW]

b)Pacific

Baren(s/Kara SeaEas(-Siberian SeaSignificant (95%)Significant (99%)

Fig. 8 Regressions of SIC in the Barents and Kara Seas (red)and Chukchi and East-Siberian Seas (blue) on the latent en-ergy transport by planetary waves across 70 ◦N over the At-lantic (a) and Pacific (b) sector. Regressions are given as afunction of time lag, in units of SIC anomaly in million km2

per PW of transport. Regression significant on a 95 and 99 %level are shown in light and dark-green shading, respectively.

the large planetary waves. Hence, a transport event up563

through the Atlantic sector causes winds and increase564

in surface energy flux over the Barents and Kara Seas,565

resulting in a negative sea-ice anomaly there, but a sig-566

nificant reduction in surface energy flux and positive567

ice anomaly over the Chukchi and East-Siberian Seas568

(Figs. 8 and 9), as likely the planetary-wave flow has569

cooled and dried on its path over the North Pole.570

571

Following again Rydsaa et al. (2021), extreme events572

for the winter season are here presented in order to573

provide a quantitative example of the transport im-574

pact: Fig. 10 shows 5-day-lag composites of sea-ice and575

surface-wind anomalies that correspond to extreme la-576

tent planetary transport events in winter. Extreme events577

are defined as days when the zonal-mean transport anoma-578

lies are larger than the 97.5th percentile. Fig. 10a is for579

all events, with maximum transport occurring at any580

longitude at 70 ◦N, whereas Fig. 10b and c present ex-581

treme events occurring in the Atlantic and the Pacific582

sector, respectively. 583

584

For extreme latent planetary wave transport through 585

the Atlantic sector, strong significant negative sea-ice 586

anomalies are found in the Barents and Kara Seas, 587

whereas positive anomalies appear south of the Bering 588

Strait (Fig. 10b). This transport pattern is accompa- 589

nied by anomalies of downwelling longwave radiation 590

showing significantly positive anomalies north of the 591

Atlantic sector towards the North Pole, but switching 592

to significantly negative anomalies in the vicinity and 593

south of the Bering Strait (Fig. 11a). In the vicinity 594

of the Barents and Kara Seas, where the marginal ice 595

zone is often encountered in winter, large part of the 596

longwave radiation is absorbed by the surface causing 597

ice melt or delaying freeze up, which is apparent from 598

the significant positive anomaly of the net longwave ra- 599

diation over this region (Fig. 11b). The positive energy 600

balance in favour of the surface in this region is en- 601

hanced by reduction of latent and sensible turbulent 602

fluxes (Fig. 11c and d). 603

604

Further to the north in the central Arctic, the sea-ice 605

anomalies are much smaller as presumably here the ice 606

is thicker and temperature lower. Hence even though 607

these transport extremes induce increase in longwave 608

radiation over the central Arctic, this excess in surface 609

energy warms the surface rather than causes significant 610

ice area changes, resulting in a corresponding increase 611

in up-welling radiation and, as a consequence, little 612

change in net radiation. Consistent with the ice show- 613

ing little change in the central Arctic, also the turbu- 614

lent surface fluxes remain little affected here during the 615

extreme transport events. South of the Bering Strait, 616

where the downwelling longwave radiation shows nega- 617

tive anomalies and the sea-ice cover increases, also the 618

net longwave radiation anomaly is negative, indicating 619

that the surface temperature stays about constant over 620

this region during the freeze up. 621

622

Also surface wind anomalies associated with the trans- 623

port extremes may play a role for the sea-ice response. 624

Over the ice-retreat area in the vicinity of the Bar- 625

ents and Kara Seas, anomalously strong winds from the 626

south appear, which may break up ice into flows and 627

cause leads and polynyas. 628

629

For the transport extremes in the Pacific sector, re- 630

duction of ice is found in the winter marginal ice zone 631

south of the Bering Strait, and consistent with the At- 632

lantic extremes, positive ice anomalies are found on the 633

opposite site of the Arctic (Fig. 10c). When it comes to 634

all extremes, their transport maxima are mostly con- 635


40 30 20 10 0 10 20 30 40

15

10

5

0

5

10

15

20

25

Regr

essio

n co

effic

ient

[W/m

2 /PW

]

a) Atlantic, LWS + SSH + SLH

Barents/Kara SeaEast-Siberian Sea

Significant (95%)Significant (99%)

40 30 20 10 0 10 20 30 40

15

10

5

0

5

10

15

20

25b) Pacific, LWS + SSH + SLH



40 30 20 10 0 10 20 30 40Time lag [days]

0.4

0.2

0.0

0.2

0.4

0.6

Regr

essio

n co

effic

ient

[m/s

/PW

]

c) Atlantic, 10m wind



40 30 20 10 0 10 20 30 40Time lag [days]

0.4

0.2

0.0

0.2

0.4

0.6

d) Pacific, 10m wind



Fig. 9 As Fig. 8, but for regressions over the (a, b) LWS+SSH+SLH, and (b, c) 10 m wind on the (a, c) Atlantic, and (b, c)Pacific sector planetary latent energy transport across 70 ◦N.

a)Total1m

s

b)Atlantic1m

sc)Pacific

1ms

-0.14

-0.09

-0.04

0.00

0.05

0.09

0.13

SIC

anom

aly

[-]

Fig. 10 Composites of SIC and wind anomalies for planetary latent transport extreme events (>97.5th percentile) at lag 5days after the event maximum for winter (December-February). Green dots along 70 ◦N indicate the location of the maximumtransport of each extreme event, based on (a) all longitudes, and over the (b) Atlantic and (c) Pacific sectors defined as from30 ◦W to 30 ◦E and from 145 ◦W to 160 ◦E, respectively. Anomalies are relative to a climatology over 1979-2018. Shadingindicates SIC anomalies, vectors show 10 m-wind anomalies in m/s, and green contours encapsulate areas where SIC anomaliesare significant on a 99% level based on a Monte-Carlo approach with 5000 simulations.


a) LWSD1m

s

b) LWS1m

s

c) SSH1m

s

d) SLH1m

s

-27.00

-18.00

-9.00

0.00

9.00

18.00

27.00

Ener

gy a

nom

aly

[W/m

2 ]Fig. 11 As Fig. 10b, but for surface (a) down-welling longwave radiation, (b) net longwave radiation, (c) latent heat flux,and (d) sensible heat flux anomalies.

fined to the Atlantic and Pacific sectors, with the for-636

mer sector being the dominating (Fig. 10a). Hence, the637

composite over all extremes is dominated by the At-638

lantic pattern.639

640

In summary, extremes of latent planetary-wave trans-641

port during winter most often occur in one of the the642

two ocean sectors. The transport extremes likely lead643

to a trans-Arctic flow causing a convergence of humid-644

ity, a positive surface energy balance, and a negative645

ice anomaly at the poleward branch of the transport,646

and the opposite at the southward transport branch at647

the other side of the North Pole. In addition, also a648

dynamical effect of increased surface winds plays a role649

for the ice response just north of the Arctic entrance of650

the extreme events. 651

652

The summer season shows a similar pattern as com- 653

pared to winter with regard to negative ice anoma- 654

lies north of the sector of extreme-transport entrance, 655

and positive anomalies at the other side of the Arc- 656

tic (Fig. 12b). However in the summer, when the ice 657

is generally thinner, surface winds likely play an even 658

larger role for the ice response due to large-scale drift. 659

Extreme transport events through the Atlantic sector 660

cause cyclonic wind anomalies over the central Arctic, 661

which presumably play an important role in piling up 662

the sea ice in the Beaufort, Chukchi and East-Siberian 663

Seas (Fig. 12b). The piling-up effect down-stream of 664

the Pacific transports is seen only in the Laptev Sea 665

and is less likely to occur at the Atlantic side of the 666


a)Total1m

s

b)Atlantic1m

sc)Pacific

1ms

-0.14

-0.09

-0.04

0.00

0.05

0.09

0.13

SIC

anom

aly

[-]

Fig. 12 As Fig. 10, but for the summer season (June-August)

Arctic ocean, where more open pathways to southern667

latitudes are found, and sea ice is lost due to transport668

through the Fram strait (Fig. 12c). From Fig. 12a it is669

indicated that transport extremes in summer are less670

confined to the Atlantic and Pacific sectors, than is the671

case for the winter conditions, but occur also over the672

continents.673

674

4 Discussion675

Planetary and synoptic-scale waves are strongly char-676

acterizing the weather for many of us living in the mid-677

latitudes. Whereas cyclones in many cases bring windy678

and rainy weather that often last for an afternoon or a679

day, the large-scale waves are associated with warm or680

cold and rainy or sunny conditions sustaining for weeks,681

for instance associated with blocking events. The plan-682

etary waves induced by e.g. topography and land-sea683

heating contrasts meander with varying amplitude and684

zonal phase speed bringing warm and humid air north-685

ward at some latitudes and cold and dry air southward686

at others. The large-scale nature of these waves implies687

that a northward flank during strong events can reach688

large parts of the Arctic where wave breaking in form689

of convergence of heat and moisture can take place.690

Cyclones on the other hand are smaller in nature and691

induced by baroclinic instability where the large-scale692

meriodional temperature gradient is an important driv-693

ing force. Hence strong cyclones event predominantly694

occur during cold Arctic conditions, and their Arctic695

impact is limited by their spatial scales.696

Baggett and Lee (2015) presented interesting work697

along these lines and showed based on the Lorenz en-698

ergy cycle frame work that while strong eddy kinetic699

energy on the synoptic scale is proceeded by enhanced700

zonal-mean available potential energy, consistent with 701

a strong meriodional temperature gradient, the eddy- 702

kinetic energy on the planetary scale shows no such de- 703

pendence. Both types of waves lead to a decrease in the 704

zonal-mean available potential energy, consistent with 705

Arctic warming. They also found that the Arctic warm- 706

ing effect is considerably larger for the planetary than 707

for the synoptic waves. These findings are supported by 708

Graversen and Burtu (2016) and Rydsaa et al. (2021), 709

and are consistent with our results suggesting that plan- 710

etary waves have a much stronger effect on Arctic sea 711

ice than have cyclones for the same amount of energy 712

transported by the two types of waves. Other studies 713

have earlier pointed to the importance of cyclones and 714

change in “storm tracks” when it comes to the impact 715

on Arctic climate and weather from atmospheric circu- 716

lation (Woods et al., 2013; Shaw et al., 2016; Schlichtholz, 717

2018; Nygard et al., 2019). However, this does not nec- 718

essarily imply a discrepancy between these studies and 719

our results as well as other studies arguing for the larger 720

importance of planetary waves as compared to those of 721

the synoptic scale type: “Storm tracks” which can be 722

regarded as a cyclone wave guide may be just part of 723

Rossby wave. The baroclinic zone at the boundary be- 724

tween polar and subtropical air masses also constitutes 725

the Rossby waves, and cyclones generated here are ad- 726

vected by these planetary waves. Hence many of the 727

phenomena discussed in the studies mentioned above, 728

arguing for the importance of the cyclones, may be coin- 729

ciding with, and part of planetary-wave transport. This 730

is in agreement with Papritz and Dunn-Sigouin (2020) 731

finding that cyclones play a large role for the moisture 732

transport to the Arctic, but that this transport is found 733

on the planetary scales since blockings on these scales 734

act as a wave guide for the cyclones. 735


Consistent with earlier studies, we find that latent736

transport as compared to the dry-static has a much737

larger effect on Arctic climate for the same amount of738

energy transported (Woods et al., 2013; Graversen and739

Langen, 2019). However the variability of the dry-static740

transport is an order of magnitude larger than that of741

the latent, implying that the two types of energy trans-742

port show impact on the sea ice of similar magnitude.743

This is consistent with Olonscheck et al. (2019) finding744

that moist-static energy is important for Arctic sea-ice745

variability. However when it comes to model estimation746

of future trends of Arctic impact by circulation changes,747

Graversen and Langen (2019) showed that although the748

dry-static transport decreases more than the latent in-749

creases, the total response is still Arctic warming due to750

a stronger warming potential of the latent as compared751

to the dry-static component.752

5 Conclusion753

In this study we show the impact of latent as well as754

dry-static atmospheric energy transport on the Arctic755

sea ice cover. By analysing regressions of sea-ice anoma-756

lies on anomalies in the transport components of both757

planetary and synoptic scale waves we come to the fol-758

lowing conclusions:759

– Latent planetary energy transport results in a stronger760

decrease in sea ice per unit of energy transported as761

compared to the dry-static component of the trans-762

port;763

– Due to the larger variability in dry-static planetary764

transport, the actual impact of dry-static transport765

on the Arctic sea ice is comparable in magnitude as766

compared to latent transport;767

– The impact of synoptic-scale waves on the Arctic sea768

ice is considerably smaller than that of the planetary769

waves. Positive regressions indicate that synoptic-770

scale waves occur during cold Arctic conditions;771

– Winter extreme transport events occur often either772

over the Atlantic or over the Pacific sector, resulting773

in decreased ice conditions in the nearby regions and774

increased ice conditions in regions on the opposite775

site of the Arctic.776

Funding777

This work was funded by the The Research Council778

of Norway (NFR) under the project “The role of the779

atmospheric energy transport in recent Arctic climate780

change” (no. 280727).781

Conflicts of interests 782

The authors have no conflicts of interest to declare that 783

are relevant to the content of this article. 784

Code and data availability 785

Data and scripts used for this work are stored at the 786

Nird storage facility provided by the Norwegian e-infrastructure787

for research and education, UNINETT Sigma2. Data 788

can be accessed on request to the corresponding au- 789

thor. 790

Authors’ contribution 791

M. G. Hofsteenge performed most of the data process- 792

ing and analyses. R. G. Graversen calculated the ERA5 793

energy transport. M. G. Hofsteenge and R. G. Gra- 794

versen wrote the paper with input from all co-authors. 795

All authors discussed results and contributed with ideas. 796

Acknowledgements We thank the ERA5 development group 797

for providing the publicly accessible data on the ECMWF 798

server that was used in this study. The data was processes on 799

the Stallo and Fram supercomputer and stored at the Nird 800

storage facilities provided by the Norwegian e-infrastructure 801

for research and education, UNINETT Sigma2, under the 802

projects NN9348K and NS9063K. This work was funded by 803

the The Research Council of Norway (NFR) under the project 804

“The role of the atmospheric energy transport in recent Arc- 805

tic climate change” (no. 280727). 806

References 807

Arthun M, Eldevik T, Smedsrud LH, Skagseth, Ingvald- 808

sen RB (2012) Quantifying the influence of Atlantic 809

heat on Barents sea ice variability and retreat. Jour- 810

nal of Climate 25(13):4736–4743, DOI 10.1175/JCLI- 811

D-11-00466.1 812

Baggett C, Lee S (2015) Arctic warming induced by 813

tropically forced tapping of available potential energy 814

and the role of the planetary-scale waves. Journal 815

of the Atmospheric Sciences 72(4):1562–1568, DOI 816

10.1175/JAS-D-14-0334.1 817

Baggett C, Lee S, Feldstein S (2016) An investigation of 818

the presence of atmospheric rivers over the North Pa- 819

cific during planetary-scale wave life cycles and their 820

role in Arctic warming. Journal of the Atmospheric 821

Sciences 73(11):4329–4347, DOI 10.1175/JAS-D-16- 822

0033.1 823

Boisvert LN, Petty AA, Stroeve JC (2016) The im- 824

pact of the extreme winter 2015/16 Arctic cyclone 825


on the Barents-Kara Seas. Monthly Weather Review826

144, DOI 10.1175/MWR-D-16-0234.s1827

Doscher R, Koenigk T (2013) Arctic rapid sea ice loss828

events in regional coupled climate scenario experi-829

ments. Ocean Science 9(2):217–248, DOI 10.5194/os-830

9-217-2013831

Fox-Kemper B, Hewitt H, Xiao C, Aalgeirsdottir G,832

Drijfhout S, Edwards T, Golledge N, Hemer M,833

Kopp R, Krinner G, Mix A, Notz D, Nowicki S,834

Nurhati S, Ruiz L, Sallee JB, Slangen A, Yu Y,835

R E Kopp GK (2021) Ocean, Cryosphere and Sea836

Level Change. In: MassonDelmotte V, Zhai P, Pirani837

A, Connors SL, Pean C, Berger S, Caud N, Chen Y,838

Goldfarb L, Gomis MI, Huang M, Leitzell K, Lon-839

noy E, Matthews JBR, Maycock TK, Waterfield T,840

Yelekci O, Yu R, Zhou B (eds) Climate Change 2021:841

The Physical Science Basis. Contribution of Working842

Group I to the Sixth Assessment Report of the Inter-843

governmental Panel on Climate Change, Cambridge844

University Press845

Graham RM, Itkin P, Meyer A, Sundfjord A, Spreen G,846

Smedsrud LH, Liston GE, Cheng B, Cohen L, Divine847

D, Fer I, Fransson A, Gerland S, Haapala J, Hudson848

SR, Johansson AM, King J, Merkouriadi I, Peter-849

son AK, Provost C, Randelhoff A, Rinke A, Rosel A,850

Sennechael N, Walden VP, Duarte P, Assmy P, Steen851

H, Granskog MA (2019) Winter storms accelerate the852

demise of sea ice in the Atlantic sector of the Arctic853

Ocean. Scientific Reports 9(1), DOI 10.1038/s41598-854

019-45574-5855

Graversen RG (2006) Do changes in the midlatitude856

circulation have any impact on the Arctic surface air857

temperature trend? Journal of Climate 19:5422–5438858

Graversen RG, Burtu M (2016) Arctic amplification en-859

hanced by latent energy transport of atmospheric860

planetary waves. Quarterly Journal of the Royal861

Meteorological Society 142(698):2046–2054, DOI862

10.1002/qj.2802863

Graversen RG, Langen PL (2019) On the role of the864

atmospheric energy transport in 2×3CO2-induced865

polar amplification in CESM1. Journal of Climate866

32(13):3941–3956, DOI 10.1175/JCLI-D-18-0546.1867

Graversen RG, Mauritsen T, Drijfhout S, Tjernstrom868

M, Martensson S (2011) Warm winds from the Pa-869

cific caused extensive Arctic sea-ice melt in summer870

2007. Climate Dynamics 36(11-12):2103–2112, DOI871

10.1007/s00382-010-0809-z872

Hegyi BM, Taylor PC (2018) The unprecedented873

2016–2017 Arctic sea ice growth season: The crucial874

role of atmospheric rivers and longwave fluxes. Geo-875

physical Research Letters 45(10):5204–5212, DOI876

10.1029/2017GL076717877

Heiskanen T, Graversen RG, Rydsaa JH, Isachsen PE 878

(2020) Comparing wavelet and Fourier perspectives 879

on the decomposition of meridional energy transport 880

into synoptic and planetary components. Quarterly 881

Journal of the Royal Meteorological Society DOI 882

10.1002/qj.3813 883

Hersbach H, Bell B, Berrisford P, Hirahara S, Horanyi 884

A, Munoz-Sabater J, Nicolas J, Peubey C, Radu R, 885

Schepers D, Simmons A, Soci C, Abdalla S, Abel- 886

lan X, Balsamo G, Bechtold P, Biavati G, Bidlot 887

J, Bonavita M, De Chiara G, Dahlgren P, Dee D, 888

Diamantakis M, Dragani R, Flemming J, Forbes R, 889

Fuentes M, Geer A, Haimberger L, Healy S, Hogan 890

RJ, Holm E, Janiskova M, Keeley S, Laloyaux P, 891

Lopez P, Lupu C, Radnoti G, de Rosnay P, Rozum 892

I, Vamborg F, Villaume S, Thepaut JN (2020) The 893

ERA5 global reanalysis. Quarterly Journal of the 894

Royal Meteorological Society 146(730):1999–2049, 895

DOI 10.1002/qj.3803 896

Hirahara S, Alonso-Balmaseda M, de Boisseson E, 897

Hersbach H (2016) Sea surface temperature and sea 898

ice concentration for ERA5 — ECMWF. In: ERA 899

Report Series No. 26 900

Kapsch ML, Graversen RG, Tjernstrom M (2013) 901

Springtime atmospheric energy transport and the 902

control of Arctic summer sea-ice extent. Nature Cli- 903

mate Change DOI 10.1038/nclimate1884 904

Kapsch ML, Graversen RG, Tjernstrom M, Bintanja R 905

(2016) The effect of downwelling longwave and short- 906

wave radiation on Arctic summer sea ice. Journal of 907

Climate 29(3):1143–1159, DOI 10.1175/JCLI-D-15- 908

0238.1 909

Kapsch ML, Skific N, Graversen RG, Tjernstrom M, 910

Francis JA (2019) Summers with low Arctic sea ice 911

linked to persistence of spring atmospheric circula- 912

tion patterns. Climate Dynamics 52(3-4):2497–2512, 913

DOI 10.1007/s00382-018-4279-z 914

Kashiwase H, Ohshima KI, Nihashi S, Eicken H (2017) 915

Evidence for ice-ocean albedo feedback in the Arc- 916

tic Ocean shifting to a seasonal ice zone. Scientific 917

Reports 7(1), DOI 10.1038/s41598-017-08467-z 918

Kay JE, L’Ecuyer T (2013) Observational constraints 919

on Arctic Ocean clouds and radiative fluxes dur- 920

ing the early 21st century. Journal of Geophysi- 921

cal Research Atmospheres 118(13):7219–7236, DOI 922

10.1002/jgrd.50489 923

Kay JE, L’Ecuyer T, Gettelman A, Stephens G, O’Dell 924

C (2008) The contribution of cloud and radiation 925

anomalies to the 2007 Arctic sea ice extent min- 926

imum. Geophysical Research Letters 35(8), DOI 927

10.1029/2008GL033451 928

Liu Z, Schweiger A (2017) Synoptic conditions, clouds, 929

and sea ice melt onset in the Beaufort and Chukchi 930


seasonal ice zone. Journal of Climate 30(17):6999–931

7016, DOI 10.1175/JCLI-D-16-0887.1932

Mills CM, Walsh JE (2014) Synoptic activity asso-933

ciated with sea ice variability in the Arctic. Jour-934

nal of Geophysical Research 119(21):117–12, DOI935

10.1002/2014JD021604936

Mortin J, Svensson G, Graversen RG, Kapsch ML,937

Stroeve JC, Boisvert LN (2016) Melt onset over Arc-938

tic sea ice controlled by atmospheric moisture trans-939

port. Geophysical Research Letters 43(12):6636–940

6642, DOI 10.1002/2016GL069330941

Nygard T, Graversen RG, Uotila P, Naakka T, Vihma T942

(2019) Strong dependence of wintertime Arctic mois-943

ture and cloud distributions on atmospheric large-944

scale circulation. Journal of Climate 32(24):8771–945

8790, DOI 10.1175/JCLI-D-19-0242.1946

Ogi M, Yamazaki K, Wallace JM (2010) Influence of947

winter and summer surface wind anomalies on sum-948

mer Arctic sea ice extent. Geophysical Research Let-949

ters 37(7):–, DOI 10.1029/2009GL042356950

Olonscheck D, Mauritsen T, Notz D (2019) Arc-951

tic sea-ice variability is primarily driven by atmo-952

spheric temperature fluctuations. Nature Geoscience953

12(6):430–434, DOI 10.1038/s41561-019-0363-1954

Onarheim IH, Eldevik T, Smedsrud LH, Stroeve JC955

(2018) Seasonal and regional manifestation of Arc-956

tic sea ice loss. Journal of Climate 31(12):4917–4932,957

DOI 10.1175/JCLI-D-17-0427.1958

Papritz L, Dunn-Sigouin E (2020) What configura-959

tion of the atmospheric circulation drives extreme960

net and total moisture transport into the Arc-961

tic. Geophysical Research Letters 47(17), DOI962

10.1029/2020GL089769963

Park HS, Lee S, Feldstein SB, Kosaka YU (2015) The964

impact of poleward moisture and sensible heat flux on965

Arctic winter sea ice variability. Journal of Climate966

28:5030–5040, DOI 10.1175/JCLI-D-15967

Parkinson CL, Cavalieri DJ (2008) Arctic sea ice968

variability and trends, 1979-2006. Journal of Geo-969

physical Research: Oceans 113(7):7003, DOI970

10.1029/2007JC004558971

Peng G, Meier WN (2018) Temporal and regional972

variability of Arctic sea-ice coverage from satellite973

data. Annals of Glaciology 59(76pt2):191–200, DOI974

10.1017/aog.2017.32975

Rydsaa J, Graversen R, Heiskanen TIH, Stoll PJ (2021)976

Changes in atmospheric latent energy transport into977

the Arctic; planetary versus synoptic scales. Quar-978

terly Journal of the Royal Meteorological Society p979

qj.4022, DOI 10.1002/qj.4022980

Schlichtholz P (2018) Climate impacts and Arctic981

precursors of changing storm track activity in the982

Atlantic-Eurasian region. Scientific Reports 2018 8:1983

8(1):1–19, DOI 10.1038/s41598-018-35900-8 984

Schweiger AJ, Zhang J, Lindsay RW, Steele M (2008) 985

Did unusually sunny skies help drive the record sea 986

ice minimum of 2007? Geophysical Research Letters 987

35(10), DOI 10.1029/2008GL033463 988

Screen JA, Simmonds I, Keay K (2011) Dramatic in- 989

terannual changes of perennial Arctic sea ice linked 990

to abnormal summer storm activity. Journal of 991

Geophysical Research Atmospheres 116(15), DOI 992

10.1029/2011JD015847 993

Serreze MC, Stroeve J (2015) Arctic sea ice trends, 994

variability and implications for seasonal ice forecast- 995

ing. Philosophical Transactions of the Royal Society 996

A: Mathematical, Physical and Engineering Sciences 997

373(2045), DOI 10.1098/rsta.2014.0159 998

Serreze MC, Box JE, Barry RG, Walsh JE (1993) Char- 999

acteristics of Arctic synoptic activity, 1952-1989. Me- 1000

teorology and Atmospheric Physics 51(3-4):147–164, 1001

DOI 10.1007/BF01030491 1002

Shaw TA, Baldwin M, Barnes EA, Caballero R, 1003

Garfinkel CI, Hwang YT, Li C, O’Gorman PA, 1004

Riviere G, Simpson IR, Voigt A (2016) Storm track 1005

processes and the opposing influences of climate 1006

change. Nature Geoscience 2016 9:9 9(9):656–664, 1007

DOI 10.1038/ngeo2783 1008

Simmonds I, Rudeva I (2012) The great Arctic cyclone 1009

of August 2012. Geophysical Research Letters 39(23), 1010

DOI 10.1029/2012GL054259 1011

Trenberth KE (1991) Climate diagnostics from global 1012

analyses: Conservation of mass in ECMWF analyses. 1013

Journal of Climate 4(7) 1014

Wang Z, Walsh J, Szymborski S, Peng M (2020) Rapid 1015

Arctic sea ice loss on the synoptic time scale and 1016

related atmospheric circulation anomalies. Journal of 1017

Climate 33:1597–1617, DOI 10.1175/JCLI-D-19 1018

Woods C, Caballero R (2016) The role of moist in- 1019

trusions in winter arctic warming and sea ice de- 1020

cline. Journal of Climate 29(12):4473–4485, DOI 1021

10.1175/JCLI-D-15-0773.1 1022

Woods C, Caballero R, Svensson G (2013) Large-scale 1023

circulation associated with moisture intrusions into 1024

the Arctic during winter. Geophysical Research Let- 1025

ters 40(17):4717–4721, DOI 10.1002/grl.50912 1026

Yang W, Magnusdottir G (2017) Springtime extremem- 1027

oisture transport into the Arctic and its impact on sea 1028

ice concentration. Journal of Geophysical Research 1029

122(10):5316–5329, DOI 10.1002/2016JD026324 1030

Zhang J, Lindsay R, Schweiger A, Steele M (2013) 1031

The impact of an intense summer cyclone on 2012 1032

Arctic sea ice retreat. Geophysical Research Letters 1033

40(4):720–726, DOI 10.1002/grl.50190 1034

Zhang X, Walsh JE, Zhang J, Bhatt US, 1035

Ikeda M (2004) Climatology and interan- 1036


nual variability of Arctic cyclone activity:1037

1948-2002. Journal of Climate 17(12):2300–1038

2317, DOI https://doi.org/10.1175/1520-1039

0442(2004)017¡2300:CAIVOA¿2.0.CO;21040

Documents

The impact of atmospheric Rossby waves and cyclones on …