Dynamics and Variability of the Circulation in the North-Atlantic Subpolar Seas

Dynamics and Variability of the Circulation in the North-Atlantic Subpolar Seas

Léon Chafik

Cover image: The North Atlantic circulation including sea surface temperature and clouds.

Picture made using Google Earth.

c

Léon Chafik, Stockholm 2014

ISBN 978-91-7447-840-2

Printed in Sweden by US-AB, Stockholm 2014

Distributor: Department of Meteorology, Stockholm University

Abstract

This thesis deals with the dynamics and circulation in the northern North At-

lantic and the Nordic Seas, processes of crucial importance for the mild climate

of Scandinavia and Northern Europe. High-resolution ADCP scans of currents

from Greenland to Scotland in the top 400 m demonstrate that the Reykjanes

Ridge is a very effective separator of flow towards the Nordic and Labrador

Seas, respectively. It was found that the meridional overturning circulation

has weakened by ∼1.7 Sv (1 Sv = 10 ⁶ m ³ s ⁻¹ ) during the 18-year period when

altimetric data were available. This trend may be an effect of the Atlantic Mul-

tidecadal Oscillation, but is certainly not due to the North Atlantic Oscillation

(NAO). By studying the circulation in the Faroe-Shetland Channel, which is

an important choke point for the global thermohaline circulation, it was con-

cluded that the contraction of the Norwegian-Sea gyre during low NAO peri-

ods plays an important role for disturbing the flow pattern. This specifically

affects the regional ocean climate by leading to an accumulation of warm and

saline Atlantic waters in the channel. During high NAO phases the circulation

is strongly topographically controlled. The Norwegian Atlantic Slope Current

(NwASC) is the main flow branch linking the North Atlantic to the Arctic and

Barents Sea. It was found that the NwASC is largely coherent over seasonal to

interannual time-scales. However, on shorter time-scales the coherency of the

flow shows a sustained and pronounced weakening downstream of Lofoten. In-

tense eddy-shedding from the slope into the Lofoten Basin damps the coherent

structure of the flow. The eddies take about two months to propagate to and to

merge with the semi-permanent anticyclonic vortex above the deepest part of

the Lofoten Basin. These results have implications for how flow/hydrographic

anomalies are transferred through the Nordic Seas towards the Arctic. Anoma-

lous transports of warm water into the Arctic and Barents Sea via the NwASC

are found to be driven by a combination of the NAO and the other two leading

modes of atmospheric variability in the North Atlantic. The results reported in

the thesis may be of importance for achieving a correct representation of the

heat conveyed polewards in climate models.

This thesis is dedicated to my beloved grandmother,

who took care of me since the instant I was born,

and made me who I am today.

After all of this time After all of these seasons After your one decision to go to the water for reason It’s only the ocean and you And all of these lines will all be erased soon They go out with the tide Then come back with the waves It’s only the ocean and you

From ”Only the ocean” by Jack Johnson

List of Papers

The following papers, referred to in the text by their Roman numerals, are included in this thesis.

PAPER I: On the spatial structure and temporal variability of pole- ward transport between Scotland and Greenland

Chafik L., Rossby T. and Schrum C., J. Geophys. Res., 119 (2014).

DOI: 10.1002/2013JC009287

PAPER II: The response of the circulation in the Faroe-Shetland Chan- nel to the North Atlantic Oscillation

Chafik L. Tellus A, 64, 18423 (2012).

DOI: tellusa.v64i0.18423

PAPER III: Transport coherency and Eddy activity along the Norwegian- Atlantic Slope Current

Chafik L., Nilsson J., Skagseth, Ø, Lundberg P., Submitted to J.

Phys. Oceanogr.

PAPER IV: The Lofoten Vortex of the Nordic Seas

Raj P. R., Chafik L., Nilsen J. E. Ø. and Eldevik T., In review Deep Sea Res.

PAPER V: Atmospheric circulation patterns control the variability of the oceanic transport towards the Arctic region

Chafik L., Nilsson J., Skagseth Ø., Lundberg P. and Hannachi A., Manuscript.

Papers I and II reproduced with permission from the American Geo-

physical Union and Tellus A.

Papers not included in this thesis

• Excitation of equatorial Kelvin and Yanai waves by tropical cyclones in an ocean general circulation model

Sriver R. L., Huber M., Chafik L., Earth Syst. Dynm., 4, 1-10 (2013).

DOI: 10.5194/esd-4-1-2013

• Estimates of the Southern Ocean general circulation improved by animal-borne instruments

Fabien Roquet, Carl Wunsch, Gael Forget, Patrick Heimbach, Christophe Guinet, Gilles Reverdin, Jean-Benoit Charrassin, Frederic Bailleul, Daniel P. Costa6, Luis A. Huckstadt, Kimberly T. Goetz, Kit M. Kovacs, Chris- tian Lydersen, Martin Biuw, Ole A. Nøst, Horst Bornemann, Joachim Ploetz, Marthan N. Bester, Trevor McIntyre, Monica C. Muelbert, Mark A. Hindell, Clive R. McMahon, Guy Williams, Robert Harcourt, Iain C.

Field, Léon Chafik, Keith W. Nicholls, Lars Boehme, Mike A. Fedak

Geophys. Res. Lett., 40, 6176-6180 (2013).

Author’s contribution

The idea of comparing ADCP and altimetry in paper I originated from a dis- cussion between the authors. I performed the analysis with the aid of T.

Rossby. The paper was written together with T. Rossby and C. Schrum. Paper II is my own idea. I collected all data, analysed it and written the paper. Con- structive comments were received from Peter Lundberg and Johan Nilsson.

For paper III, the idea came from discussions between me, Johan Nilsson,

Øystein Skagseth and Peter Lundberg. I conducted the analysis and the paper

was written jointly. In paper IV, the data analysis as well as the writing of the

slope current part is done by me. The ideas underlying paper V are mine and

I performed all the analysis. The paper was written with inputs from Johan

Nilsson, Øystein Skagseth, Peter Lundberg and Abdel Hannachi.

Contents

Abstract iii

List of Papers vii

Author’s contribution ix

1 Introduction 13

2 Satellite altimetry, oceanic observations and methodology 15

2.1 Satellite altimetry . . . . 15

2.1.1 Principles of satellite altimetry . . . . 15

2.1.2 Currents, eddies and their interaction . . . . 16

2.2 Hydrostatics and geostrophic relations . . . . 19

2.3 Estimating the volume transport . . . . 21

2.4 Vessel-mounted ADCP . . . . 22

3 The circulation of the Subpolar Seas 25 3.1 The circulation of the northern North Atlantic . . . . 25

3.2 The circulation of the Nordic Seas . . . . 27

4 The Atlantic Oscillations 31 4.1 The North Atlantic Oscillation . . . . 31

4.1.1 The Scandinavian and East Atlantic patterns . . . . 33

4.2 The Atlantic Multidecadal Oscillation . . . . 35

5 Outlook 37

Sammanfattning xxxix

Acknowledgements xli

References xliii

1. Introduction

The mild climate of Northern Europe and Scandinavia owes its existence to the poleward transport of warm and saline Atlantic waters towards the north- ern North Atlantic and into Nordic Seas. This heat is conveyed to higher lati- tudes by the Gulf Stream and the North Atlantic Current (NAC) systems, which constitute an important part of the upper limb of the Atlantic Meridional Over- turning Circulation (AMOC) and thus play a critical role in the climate system.

The first explorer who discovered that the North Atlantic Current extends into the Nordic Seas was perhaps Martin Frobisher on his voyage north-west from Ireland. He subsequently wrote:

”...with a great currante from oute of the south-west, which carryed us (by our reckoning) one point to the north-eastwardes of our said course, which currant seemed to us to continue itselfe towards Norway and other the north- east partes of the worlds...”

But the connection between the ocean circulation at high latitudes and global climate has not always been as obvious as it is nowadays, and perhaps the first oceanographer to point this out was Fridtjof Nansen, who already in 1902 stated that

”it is evident that the oceanographical conditions of the North Polar Basin have much influence upon the climate, and it is equally evident that changes in the condition of circulation would greatly change the climate conditions”

The arrival of satellite altimetry, and in particular the possibility of reg-

ularly obtaining precise global sea-surface heights and hence observations of

ocean current systems, has, among other things, resolved one of the longest-

standing problems in physical oceanography, namely that of geostrophic deter-

minations of the velocity field from hydrographic data, by obviating the need

for more-or-less arbitrary choices of ”levels of no motion”. Satellite altime-

try is also used in weather/ocean forecasting systems as well as for climate-

research applications (Fig. 1.1). It provides global trends of the sea level and

can detect the advance of the El Niño and La Niña, phenomena that have great

13

socioeconomic impacts such as flooding and droughts. Satellite altimetry has proved to be invaluable for hurricane forecasts as well as during oil-spill events by helping to minimize the environmental impacts.

In this thesis we take advantage of satellite altimetry, and, whenever possible, combine it with traditional ocean observations to study the dynamics and vari- ability of the ocean circulation in the Subpolar Seas, viz. the northern North Atlantic and the Nordic Seas.

a

b

c

Figure 1.1: Different applications of satellite altimetry: a) Sea-level trends be- tween October 1992 - January 2008. Change of sea-surface height is considered to be a primary indicator of global climate change. b) Predictions of El Niño. c) Hurricane predictions and the warm core eddy that intensified hurricane Katrina.

Figures courtesy of CLS/Cnes (http://www.aviso.oceanobs.com/) and NASA.

14

2. Satellite altimetry, oceanic observations and methodology

Since the beginning of the era of satellite altimetry, i.e. a radar altimeter in space measuring the sea surface height, our understanding of the large-scale ocean circulation and its variability, mesoscale eddies as well as global sea- level trends has been revolutionized. In this section we introduce the principles of satellite-borne radar altimetry, and we present the mean circulation and eddy activity in the northern North Atlantic and the Nordic Seas (viz. the Subpolar Seas) as inferred from altimetry. In addition, we show how this technique can be useful for estimating the volume transport along the Norwegian shelf slope. In the last subsection, we briefly introduce the measurements based on an Acoustic Doppler Current Profiler (ADCP) mounted on the merchant vessel Nuka Arctica.

2.1 Satellite altimetry

2.1.1 Principles of satellite altimetry

Basically, a radar altimeter from space measures, similar to any other radar installation, the round-trip travel time it takes for the microwave pulse to be transmitted from the satellite to the sea surface and back to the radar altimeter as illustrated in Fig. 2.1. Beside the need for precise knowledge of the satel- lite position relative to the earth’s reference ellipsoid, a range of well-known instrumental and geophysical corrections (Inverse Barometer + Sea State Bias + Ionospheric correction + Tides + Wet/Dry Tropospheric Correction) need to be taken into account in order to extract the Sea Surface Height (SSH). By subtracting the SSH from the Mean Sea Surface (MSS), i.e. the sum of the geoid and the Mean Dynamic Topography (MDT), we obtain the Sea Level Anomaly (SLA). Finally, the Absolute Dynamic Topography (ADT), which is the dynamical signal used for studying the global ocean circulation, is con- structed by adding the SLAs to the MDT. Fig. 2.2 shows an example of these quantities.

15

Satellite orbit

Mean sea surface (MSS)

Reference ellipsoid Geoid

SSH

ADT SLA

MDT

(MSS-geoid) Absolute dynamic topography (ADT)

Mean dynamic topography (MDT) Sea level anomaly (SLA) +

=

Figure 2.1: Schematic explaining the principle of satellite altimetry.

Obtaining accurate ADT observations is a major challenge due to the er- rors and inaccuracies of the the MDT. Recall from Fig. 2.1, that the MDT is the difference between the well-known MSS [Wunsch and Stammer, 1998]

and the geoid, i.e. a surface of constant potential energy relative the Earth’s gravity field, assuming no tides, ocean currents and waves etc. Determining the geoid presents a considerable challenge. However, with the introduction of satellite gravimetric estimates using GRACE (Gravity Recovery And Climate Experiment ), the MDT accuracy has improved considerably.

2.1.2 Currents, eddies and their interaction

The globally gridded (1/3 ^◦ x1/3 ^◦ ) absolute dynamic topography, provided by AVISO (Archiving, Validation and Interpretation of Satellite Oceanographic Data), is obtained on the basis of multiple satellites ”working together” to pro- vide as accurate as possible SSH signals [Ducet et al., 2000; Pascual et al., 2006]. Based on the geostrophic assumption, the ADT can be used to derive zonal (u g ) and meridional (v g ) geostrophic velocity components:

u _g = − g f

∂ ADT

∂ y , (2.1)

v _g = g f

∂ ADT

∂ x , (2.2)

where g is the gravitational acceleration and f the Coriolis parameter. The

almost 20-year time-mean circulation in the northern North Atlantic and the

16

45^oW 30^oW 15^oW 0^o 15^oE 54^oN

60^oN 66^oN 72^oN 78^oN

ADT

−70

−60

−50

−40

−30

−20

−10 0

45^oW 30^oW 15^oW 0^o 15^oE

54^oN 60^oN 66^oN 72^oN 78^oN

SLA

−20

−15

−10

−5 0 5 10

45^oW 30^oW 15^oW 0^o 15^oE

54^oN 60^oN 66^oN 72^oN 78^oN

MDT

−70

−60

−50

−40

−30

−20

−10 0

a

b

c

cm

cm

cm

Figure 2.2: a) A snapshot of the sea-level anomalies (SLAs), b) the time- invariant Mean Dynamic Topography (MDT) and c) the Absolute Dynamic To- pography (SLAs + MDT).

Nordic Seas based on these geostrophic velocities is exemplified in Fig. 2.3.

The ocean is a highly turbulent system filled with eddies or swirls (with sizes on the order of tens to hundreds of kilometres), which are mainly gen- erated through baroclinic instability. It was not until the advent of satellite altimetry that we started to gain new knowledge of their dynamics, structure, variability as well as preferred pathways of propagation. Before altimetry be- came possible, and because of the difficulty and expense in deploying current- meters and hydrographic instruments, these eddy properties were almost im- possible to ascertain.

Mesoscale eddies play an important role, analogous to the synoptic sys- tems of the atmosphere, in the global heat transport and especially across strong zonal currents such as the Antarctic Circumpolar Current. They also tend to carry salt, carbon, plankton and nutrients as they propagate, and can impact the circulation by, e.g, exchanging momentum and energy with the mean flow.

The strongest and most energetic flows in the world ocean are the west-

ern boundary currents, such as the Gulf Stream/North Atlantic current and the

17

45^oW 30^oW 15^oW 0^o 15^oE 54^oN

60^oN 66^oN 72^oN 78^oN

ABSOLUTE GEOSTROPHIC VELOCITIES (1992 2012)

Figure 2.3: The mean circulation (1992-2012) in the Subpolar Seas as deduced from geostrophic velocities based on ADT data.

Agulhas as well as Kuroshio currents. The Nordic Seas is an eddy-rich basin, where eddies dominate the entire warm and buoyant Norwegian Atlantic Cur- rent, but tend to be especially intensified in the Lofoten Basin. Our findings (papers III and IV) suggest that these eddies originate from the shelf-edge cur- rent.

When using altimeter data, the oceanic eddy field is typically mapped as the eddy kinetic energy (EKE). In Fig. 2.4, we show the climatological EKE from altimetry in the Subpolar Seas. This is based on the departures of the zonal and meridional velocities from a long-term mean, and can be formulated as

EKE = 1

2 (u ⁰² _g + v ⁰² _g ), (2.3) where

u ⁰ _g = − g f

∂ SLA

∂ y , (2.4)

v ⁰ _g = g f

∂ SLA

∂ x . (2.5)

18

The 20-year time-mean EKE in the northern North Atlantic and the Nordic Seas based on equations (2.3-5) is shown in Fig. 2.4. An interesting aspect of this EKE field is that it comprises both the high and low-frequency variability associated with eddies and currents, respectively. In paper III, we show that it is worthwhile to separate between these frequencies in order to identify the spatial patterns associated with each of them.

0 20 40 60 80 100 120 140 160 180 200

45^oW 30^oW 15^oW 0^o 15^oE

54^oN 60^oN 66^oN 72^oN 78^oN

EKE (1992−2012)

Figure 2.4: The mean EKE (1992-2012) in the Subpolar Seas as deduced from geostrophic velocity anomalies. Note that the EKE as defined here comprises variability from weekly-to-interannual time-scales.

2.2 Hydrostatics and geostrophic relations

The hydrostatic equation is given by

∂ φ

∂ z = b, (2.6)

where φ ≡ p ⁰ /ρ ₀ is the kinematic pressure anomaly and b the buoyancy anomaly.

Integrating vertically from the sea-surface η, the perturbation pressure be- comes

φ (x, y, z, t) = gη − Z ₀

z

b dz, (2.7)

19

where η is the free surface and b the buoyancy anomaly. Following the ap- proach of Fofonoff [1962] (see also Walin et al. [2004]), we define a bottom pressure anomaly as

φ B (x, y,t) ≡ φ (x, y, z = −H,t) = gη − Z ₀

−H b dz. (2.8)

Using this, we can write

φ (x, y, z, t) = φ B + Z z

−H

b dz. (2.9)

Thus, the surface height can be formulated as

η (x, y, t) = η S (x, y,t) + η B (x, y,t), (2.10) where we have introduced the steric height

η S (x, y,t) = g ⁻¹ Z ₀

−H b dz. (2.11)

Here H is the depth of the ocean and a barotropic surface height component is defined as

η B (x, y,t) = g ⁻¹ φ B (x, yt). (2.12) Using these results, the geostrophic velocity can be written

u = g

f k × ∇φ B + 1 f k × ∇

Z z

−H b dz. (2.13)

At the bottom the velocity is u B (x, y,t) = 1

f k × ∇φ B + b _B (x, y,t)

f k × ∇H. (2.14)

Accordingly, we have the result

u B · ∇H = k × ∇φ B · ∇H. (2.15)

An inviscid barotropic (i.e., an unstratified fluid with b = 0) flow at low Rossby numbers is non-divergent at leading order, which implies that

u B · ∇H = 0. (2.16)

In the Nordic Seas it has been known since the seminal study of Helland- Hansen and Nansen [1909] that the flow closely follows the bathymetry (see also Orvik and Niiler [2002]) implying that to lowest order φ B = φ B (H(x, y)).

20

Thus the stratification in the Nordic seas is sufficiently weak to make the sur- face currents approximately trace the isobaths.

We define a baroclinic potential energy anomaly as Φ(x, y, t) = −

Z ₀

−H

z · b dz. (2.17)

Using this, the depth-integrated flow can be written (U,V ) = 1

f



−H ∂ φ B

∂ y − ∂ Φ

∂ y , H ∂ φ B

∂ x + ∂ Φ

∂ x



. (2.18)

Fofonoff [1962] has referred to the two transport components as the bottom- velocity and the baroclinic contributions, respectively. However, as defined here the baroclinic component is not the transport due to the thermal wind ve- locity relative to zero bottom flow, and hence the term ”bottom-velocity com- ponent” is slightly misleading.

2.3 Estimating the volume transport

In papers III and IV we calculate the volume transport along the Norwegian Atlantic Slope Current (NwASC) using an isobath-following geostrophic ap- proach. From the weekly SSHs at times t, and under the assumption of i) a strongly topographically steered flow [Isachsen et al., 2003] and ii) a nearly- barotropic shelf-edge current [Orvik et al., 2001; Skagseth et al., 2004], we construct a geostrophically balanced along-isobath barotropic transport Ψ(s,t):

Ψ(s, t) = gH

f ∆ H ADT [Sv = 10 ⁶ m ³ s ⁻¹ ]. (2.19) Here s is a curvilinear coordinate orientated along the continental slope, g the gravitational acceleration, f the local Coriolis parameter, H=700 m the arith- metic mean between the 500- and 900-m isobaths (extracted from the ETOPO- 2’ bathymetric data-set), and ∆ H ADT ≡ ADT (H _500m , s,t) − ADT (H _900m , s,t) represents the SSH difference between these two isobaths. ∆ H ADT is deter- mined by progressing along the 500-m isobath and at discrete intervals finding the nearest point on the 900-m isobath (this to estimate the slope-current flow normal to the section), whereafter the altimeter data are linearly interpolated to these locations (Fig. 2.5). When validated against long-term time series from the Svinøy and the one-year Bjørnøya current-meter data-set, this method has been shown to perform well (cf. papers III and V).

21

500m 900m

Figure 2.5: A schematic representation of how the isobath-following barotropic transport is calculated. The grey dots represent the gridded altimetric data.

Satellite altimetry is a powerful tool, and has undoubtedly advanced our un- derstanding of the general ocean circulation, its variability and much of the mesoscale eddy activity/dynamics. Satellites have, however, the drawback that they cannot look deep into the ocean, but an Acoustic Doppler Current Profiler (ADCP) is fully capable of doing so.

2.4 Vessel-mounted ADCP

An ADCP [Flagg et al., 1998] works by repeatedly transmitting acoustic pulses

of a certain frequency in four oblique directions from beneath the ship. Imme-

diately after each transmission the instrument switches to its gated-viewing

listening mode to measure the Doppler shift of the backscattered signal as a

function of time (= increasing depth). This way one obtains a profile of rela-

tive motion between the ship and the water column along each of the acoustic

beams. Moreover, under the assumption that currents at each depth are hori-

zontal and non-divergent, the along-beam velocity components are remapped

into horizontal and vertical components relative to the ship. To obtain the ab-

solute ocean current velocity, the ship’s movement and heading as determined

from GPS navigation and a GPS-based compass are removed. It is also worth

mentioning that the high horizontal resolution and repeat sampling (by com-

mercial vessels) are the major strengths of vessel-mounted ADCPs. In paper

I, we have used the 1999-2002 150 kHz ADCP data from the freighter Nuka

22

Arctica along its tracks from Scotland to Greenland. These reached down to a depth of about 400 m, and the ADCP data were averaged to provide velocity profiles with a horizontal resolution of 2.5 km [Knutsen et al., 2005].

a

b c

Figure 2.6: a) The commercial vessel Nuka Arctica. b, c) The transmit- ting/receiving head of the150 kHz ADCP from different angles.

23

24

3. The circulation of the Subpolar Seas

3.1 The circulation of the northern North Atlantic

The North Atlantic Current (NAC), which is the northward extension of the Gulf Stream, is the main conveyor of warm and saline waters to higher lati- tudes (Fig. 3.1). The NAC flows parallel to the Grand Banks and turns anti- cyclonically at the Northwest Corner [Worthington, 1976] before drifting zon- ally eastwards and crosses the Mid-Atlantic Ridge (MAR) at two specific pas- sages, namely the Faraday (50-51 ^◦ N) and the Charlie-Gibbs (52-53 ^◦ N) Frac- ture Zones [Bower and von Appen, 2008], the latter being the widest (100 km) and deepest gap (>3500 m). After passing the MAR, the NAC splits into branches feeding the Irminger and Norwegian Seas, i.e. west and east of the Reykjanes Ridge (RR), respectively. These waters eventually densify in the Labrador and the Nordic Seas and thus play a key role in the global overturn- ing circulation.

The Reykjanes Ridge effectively separates the flow of the NAC into west- ern and eastern circulation regimes. The western regime consist of the Irminger- and Labrador-Sea cyclonic circulation. The current west of the RR exhibits two cores flowing northwards (paper I). It flows northward along the RR, joins the East Greenland Current (EGC) and follows the rim of the Labrador Sea to form the subpolar-gyre circulation [Brambilla and Talley, 2008]. On the other hand, east of the RR, there are two prominent branches. One is located in the Maury Channel (MC) and east of the Hatton Bank (HB), while the other is a Slope Current (SC) flowing northwards along the eastern margin of the European continent [Skagseth et al., 2004]. Both of the latter branches are to- pographically controlled and connect to the Nordic Seas with approximately equal amounts of transport (∼ 4 Sv).

In paper I, the reported high-resolution ADCP scans 1999-2002 of cur-

rents in the top 400 m show that the largest transports between Scotland and

Greenland, across the 59.5 ^◦ N latitude, are associated with sloping bottom to-

pographies. We also shed light on where along the RR waters from the eastern

25

part crosses over to the western part. In general, we find that there is a weak but rather persistent flow along the eastern flank of the RR. By examining the SSH along the RR, we note that south (north) of 60 ^◦ N there is a strong (weak) SSH gradient, which suggests that the flow that crosses over to the western side must be large (small) here. The source of this water crossing south of 60 ^◦ N appears to be linked to the NAC branch which retroflects back towards the RR.

This view is supported by Willis and Fu [2008], who also show a significant retroflection of the North Atlantic Current south of 60 ^◦ N based on a combined altimeter and Argo-float construction of the North Atlantic circulation.

36^oW 18^oW 0^o 18^oE 36^oE

54^oN 60^oN 66^oN 72^oN 78^oN 84^oN

Nordic Seas

Nw ASC Nw AFC RR

EGC

Arctic Ocean

Barents Sea Lofoten

MC HB

FSC

NA C

North Atlantic

Subpolar gyre

Norw ay Swed en Green

lan d

Iceland

SC

Figure 3.1: The near-surface circulation of the Subpolar Seas.

The North Atlantic circulation plays an important role for climate variabil-

ity. In paper I, we took the advantage of the 18-year satellite altimetric data-set

yielding an absolute dynamic topography to provide insights on the temporal

variability of the fluxes east and west of the Reykjanes Ridge. This excellent

altimetric coverage reveals that the poleward flows west and east of the RR are

strongly anticorrelated. We furthermore find that the two eastern branches also

are strongly anti-correlated, but offset in time with respect to the Labrador-

Sea branch. Remarkably, all these variations largely cancel out for the entire

Greenland-Scotland section, leaving a gradual decrease in sea level difference

from 0.62 to 0.56 m over the 1993-2011 satellite-observation period. This cor-

26

responds to a 10% (∼1.7 Sv) overall weakening of the meridional overturning circulation in the northern North Atlantic, almost equally divided between the Labrador and Nordic Sea branches.

3.2 The circulation of the Nordic Seas

The seas bounded by Greenland, Iceland and Norway are commonly termed the Nordic Seas and stand for 0.75% of the world ocean [Drange et al., 2005].

Albeit their small area, they play a major role in the global climate system [Rahmstorf , 1999]. This is due to the warm water carried polewards in the Nordic Seas (Figs. 3.2 and 3.3), commonly termed the Atlantic inflow, is an important source for the formation of dense water maintaining the cold equa- torward limb of the meridional overturning circulation.

12 11

10 9

7 6

8 5

4

3

2 1 0

45^oW 30^oW 15^oW 0^o 15^oE

54^oN 60^oN 66^oN 72^oN 78^oN

Temperature climatology at 50m

0 2 4 6 8 10 12

Figure 3.2: The World Ocean Atlas temperature climatology at 50 m depth. The grey contours depict the bathymetry with a spacing of 500 m.

The Nordic Seas are characterized by many features critically affecting the

circulation. Foremost among these is the bathymetry, which consists of gen-

tle and steep bottom slopes (e.g. the Vøring versus Lofoten areas), deep and

shallow basins (the Norwegian versus Iceland Seas) as well as oceanic ridges

extending from Iceland to Jan Mayen and up to the Fram Strait [Furevik and

Nilsen, 2005]. The bottom topography in the Nordic Seas plays an impor-

27

35.3

35.2

3535.1

34.9

35.1 35

34.9 34.8

35.1

45^oW 30^oW 15^oW 0^o 15^oE

54^oN 60^oN 66^oN 72^oN 78^oN

Salinity climatology at 50m

33 33.5 34 34.5 35 35.5

Figure 3.3: The World Ocean Atlas salinity climatology at 50 m depth. The grey contours depict the bathymetry with a spacing of 500 m.

tant role in guiding the flow of Atlantic water towards the Arctic region, but also for the eddy-mean flow interaction (papers II, III and IV). In fact, in the comprehensive study by Helland-Hansen and Nansen [1909], it was stated that

”As the configuration of the sea-bottom, even at great depths, has a very great influence upon the directions of currents and the circulation of the sea, even near its surface, it is much to be regretted that a more detailed knowledge of the topography of the bottom of the Norwegian Sea has not been acquired, as such knowledge would have been most desirable in discussing the circu- lation of this sea. It would be reasonable to suppose that many features of this circulation which may now seem puzzling, would then have been easily explained”.

In their report ”The Norwegian Sea: Its Physical Oceanography Based Upon The Norwegian Researches 1900-1904”, these pioneers presented a view of the circulation that remains remarkably valid. Blindheim and Østerhus [2005]

complimented the pathbreaking work of Helland-Hansen and Nansen by stat- ing that

”Their work described the sea in such detail and to such precision that

28

investigations during succeeding years could add little to their findings”.

The exchange of water masses between the North Atlantic and the Nordic Seas takes place across the Greenland-Scotland-Ridge (a natural sub-surface border to the Atlantic Ocean) via three inflow gateways, viz. between Iceland and Greenland, across the Iceland-Faroe Ridge (IFR), and through the Faroe- Shetland Channel (FSC). The two-branch system of the Norwegian Atlantic Current that transports warm and saline towards the Arctic region [Mork and Blindheim, 2000; Orvik and Niiler, 2002; Orvik et al., 2001; Poulain et al., 1996], namely the Norwegian Atlantic Slope Current (NwASC) and the Nor- wegian Atlantic Front Current (NwAFC), enters via the IFR and the FSC, re- spectively. These branches are driven by thermohaline processes and wind forcing. According to Mork and Skagseth [2010], the NwASC (3.3 Sv) trans- ports twice as much water as the NwAFC (1.7 Sv) and is thus the main pole- ward conveyor of warm Atlantic waters to the Arctic Ocean and the Barents Sea, to which it is directly connected.

The NwASC is a topographically-guided and nearly barotropic shelf-edge current [Orvik et al., 2001]. In paper III, we used satellite altimetry to re- solve the barotropic volume transport along the entire NwASC. The advantage of this approach is that it helps diagnosing where along the NwASC the co- herency, and hence mass, is lost or gained. In addition, by decomposing the along-NwASC transport into low- and high-frequency components, we were able to study its coherency on different time-scales. We find that the NwASC is largely coherent over seasonal to interannual time-scales. However, over shorter time-scales the coherency of the flow shows a pronounced weakening downstream of Lofoten. This is linked to cross-slope eddy fluxes off Lofoten, as revealed by an offshore intensification of the high-frequency eddy kinetic energy. The large-scale coherency of the NwASC has thus been related to the vigorous cross-slope shedding of eddies (Fig. 3.4) into the interior of the Lo- foten Basin (LB). These results may have implications for how oceanic signals are transferred from the North Atlantic to the Arctic. A correct representation of the eddy fluxes of Atlantic Water in the Lofoten Basin is thus a prerequisite for climate models to resolve the warm water transport to the Arctic.

Recently, the LB has received much attention [Andersson et al., 2011;

Koszalka et al., 2011; Rossby et al., 2009a,b; Søiland and Rossby, 2013; Volkov

et al., 2013], and will presumably continue to do so. In paper III, we highlight

some important aspects of the fate of Atlantic water and its connection to the

Lofoten Basin. There are, however, still many puzzling features about the dy-

namics that need to be resolved, especially concerning the quasi-permanent

29

0^o 4^oE 8^oE 12^oE 16^oE 68^oN

69^oN 70^oN 71^oN

08−Jul−2009

−40

−35

−30

−25

−20

−15

−10

−5 0 5

Figure 3.4: A snapshot (8 July 2009) of the absolute dynamic topography (cm) and its associated velocity vectors around the Lofoten Basin. The grey contours depict the bathymetry with a spacing of 500 m.

Lofoten vortex (paper IV). This anticyclone is situated over the deepest part of the LB, and it has been argued that this is a topographic β -effect [Köhl, 2007].

This vortex is characterized by anomalously high SSHs, cold sea-surface tem- peratures, high EKE levels, and a relative vorticity close to − f [Søiland and Rossby, 2013]. In paper IV, we documented its its surface and vertical charac- teristics on seasonal, inter-annual, and climatological time-scales. Moreover, our results suggest that this semi-permanent vortex over the deeper parts of the LB is maintained by eddies originating from the NwASC, which merge with and hence intensify the Lofoten vortex. Climatologically, the January transport maximum of the slope current leads the March peak in vortex mergers by two months.

30

4. The Atlantic Oscillations

4.1 The North Atlantic Oscillation

On monthy-to-interannual time scales, the circulation in the Subpolar Seas re- sponds to and can be modulated by the prevailing atmospheric circulation. In the North Atlantic, the dominating westerly winds are typically measured by the pressure difference between Iceland and the Azores, i.e. the North Atlantic Oscillation (NAO) which can be quantified by an index [Walker, 1924]. Condi- tions during a positive index are characterized by a deep Icelandic low, which is indicative of stronger-than-average westerly winds and an increase in the frequency of storm tracks towards northern Europe and Scandinavia. A nega- tive index is associated with a southward shift in the storm tracks and below- average temperatures over Europe and Scandinavia [Hurrell, 1995; Hurrell and Van Loon, 1997].

A number of oceanic phenomena in the North Atlantic have been attributed to the variability of the NAO (Fig. 4.1). In this context, one of the most in- fluential papers is that by Bjerknes [1964] on the Atlantic air-sea interaction.

By examining the relationship between the NAO and the SST in the North Atlantic, he hypothesized that the North Atlantic SST anomalies, on interan- nual time scales, are driven by changes in the atmospheric heat flux. While, on decadal time scales, ocean dynamics/circulation are more important. Most recently, this conjecture was shown to hold true, cf. Gulev et al. [2013] (see also Eden and Willebrand [2001]; Kushnir [1994]).

In the Labrador Sea, the phase of the NAO has been shown to have great influence on the convection and hence the deep water formation that feeds the global thermohaline circulation (see e.g. Curry et al. [1998]). When the NAO is high, the westerlies are strong and remove more heat from the ocean surface, a process which enhances the convective activity and produces more Labrador Sea Water (LSW). This water plays an important role for the global climate, since any reduction of the LSW, as during low NAO periods [Lazier et al., 2002; Yashayaev et al., 2007], influences the MOC.

31

Years

NAO

92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12

−3

−2

−1 0 1 2 3

75 ^o W 50 ^o W 25 ^o W 0 ^o 25 ^o E 50 ^o E 30 ^o N

40 ^o N 50 ^o N 60 ^o N 70 ^o N 80 ^o N

NAO, 39.5%

75 ^o W 50 ^o W 25 ^o W 0 ^o 25 ^o E 50 ^o E 30 ^o N

40 ^o N 50 ^o N 60 ^o N 70 ^o N 80 ^o N

SCAN, 17.5%

75 ^o W 50 ^o W 25 ^o W 0 ^o 25 ^o E 50 ^o E 30 ^o N

40 ^o N 50 ^o N 60 ^o N 70 ^o N 80 ^o N

EAP, 14.2%

Figure 4.1: The leading mode (1st EOF) of atmospheric variability in the North Atlantic, i.e. the North Atlantic Oscillation (upper panel), and its principal com- ponent (lower panel).

The strengths and shapes of the subpolar and subtropical gyres as well as their interaction are also modified by the phase of the NAO. Sarafanov [2009]

examined the modification and the spatial extent of the large-scale gyres, and showed that when the NAO is in its positive/negative phase, the subpolar gyre tends to expand/contract, which in turn leads to a retreat/advance of the sub- tropical waters. Hátún et al. [2005] highlighted that the strength and shape of the wind-driven subpolar gyre affects the inflow to the Nordic Seas. Häkki- nen et al. [2011a] underlined that a weak wind-stress curl leads to a weak- ening/contraction of the subpolar gyre but a strengthening/expansion of the subtropical gyre, a process which permits the highly saline subtropical waters to penetrate northwards, hereby presumably affecting the Nordic Seas.

32

The largest heat transport into the Nordic Seas takes place through the Faroe-Shetland Channel; an important choke-point for the global thermohaline circulation since here a significant outflow of cold deep waters as well as the inflow of warm and saline Atlantic waters take place. The volume transport of Atlantic water through the FSC has recently been estimated to be around 4±0.1 Sv [Rossby and Flagg, 2012], but a range of different transport estimates can be found in the literature (see e.g. Berx et al. [2013]; Hansen and Østerhus [2000]; Sherwin et al. [2006]). The poleward transport of North Atlantic Wa- ter (T > 9.5 ^◦ C, S > 35.3 PSU) through the FSC is mediated by the Shetland slope current, which is joined by a recirculating branch of Modified North At- lantic Water (7 < T < 9 ^◦ C, 35.1 < S < 35.3 psu) originating from north of the Faroes. Sherwin et al. [1999] showed that the coalescence of these differ- ent water masses influences the topographically steered slope current [Sherwin et al., 1999, 2006]. In paper II, we underline the importance of the large-scale wind-field for the degree of topographic control exerted on the flow in the FSC region. For a high NAO, the streamlines originating from the Norwegian Sea extend far southwards in the FSC before recirculating and merging with those of Atlantic origin. This stronger topographical steering during phases of high NAO is consistent with the anticipated increase of the wind-driven barotropic flow along the closed f /H contours (H being the depth) in the Nordic Seas.

However, under low-NAO conditions, the slope current is deflected from its path along the shelf-edge, hereby spreading comparatively warm North At- lantic waters across the FSC. This behavior is attributed to the weakened wind- stress curl inducing a contraction of the Norwegian-Sea gyre with attendant consequences for the geostrophic streamlines and hence a highly perturbed flow structure of the slope current. These results could serve as a guideline for the FSC dynamic activity to be expected during extreme NAO periods. Fac- ing the challenges of global climate change, and hence the possibility of more frequent extreme events, paper II may thus provide insights of how changing atmospheric forcing is likely to project on the FSC slope current conveying heat polewards.

4.1.1 The Scandinavian and East Atlantic patterns

The NAO is the leading mode of atmospheric variability in the North Atlantic

and European sector (Fig. 4.1), and explains 39.5% of the variance. The sec-

ond and third leading modes of variability, termed the Scandinavian (SCAN)

and East Atlantic (EAP) patterns, explain 17.5% and 14.2% of the variance,

respectively (Fig. 4.2). A positive SCAN [Bueh and Nakamura, 2007] is asso-

ciated with a positive sea-level pressure anomaly over Scandinavia, and hence

33

an anticyclonic circulation anomaly [Barnston and Livezey, 1987]. The pos- itive phase of the EAP is dominated by a well-defined low-pressure anomaly centre over the northeastern Atlantic [Murphy and Washington, 2001].

75^oW 50^oW 25^oW 0^o 25^oE 50^oE 30^oN

40^oN 50^oN 60^oN 70^oN 80^oN

NAO, 39.5%

75^oW 50^oW 25^oW 0^o 25^oE 50^oE 30^oN

40^oN 50^oN 60^oN 70^oN 80^oN

SCAN, 17.5%

75^oW 50^oW 25^oW 0^o 25^oE 50^oE 30^oN

40^oN 50^oN 60^oN 70^oN 80^oN

EAP, 14.2%

75^oW 50^oW 25^oW 0^o 25^oE 50^oE 30^oN

40^oN 50^oN 60^oN 70^oN 80^oN

NAO, 39.5%

75^oW 50^oW 25^oW 0^o 25^oE 50^oE 30^oN

40^oN 50^oN 60^oN 70^oN 80^oN

SCAN, 17.5%

75^oW 50^oW 25^oW 0^o 25^oE 50^oE 30^oN

40^oN 50^oN 60^oN 70^oN 80^oN

EAP, 14.2%

Years

EAP

92 9394 9596 97 9899 0001 02 0304 0506 07 0809 10 1112

−3

−2

−1 0 1 2 3

Years

SCAN

9293 94 9596 9798 99 0001 02 0304 0506 07 0809 1011 12

−3

−2

−1 0 1 2 3

a

c

b

d

Figure 4.2: a) The second EOF and b) third EOF of atmospheric variability in the North Atlantic, i.e. the Scandinavian (SCAN) and East Atlantic patterns (EAP), respectively. The principal component of c) SCAN and d) EAP.

Most recently, Moore et al. [2012], using several reanalysis data-sets, showed that the mobility of the MSLP centres of action in the North Atlantic are, on multidecadal time scales, influenced by the combined interaction of the NAO with the other two leading modes, i.e. the SCAN and the EAP (cf. Fig. 4.2).

Applying this type of analysis, Comas-Bru and McDermott [2013] demon- strated that the NAO alone is not enough to explain the European twentieth- century temperature and precipitation patterns.

The number of studies focusing on the oceanic response to NAO variabil- ity has increased significantly during the past two decades (see Dickson et al.

[2000], and references therein). However, studies taking into account the vari- ability of the SCAN or the EAP are still few, and, as yet, mostly based on mod- elling. Using the Bergen Climate Model [Furevik et al., 2003; Otterå et al., 2009], Langehaug et al. [2012] show that the EAP plays a role for the strength of the subpolar gyre on decadal time scales (see also Msadek and Frankig- noul [2009]). Medhaug et al. [2012] found that the NAO is of importance for the convection in the Subpolar Seas, while the poleward heat transport across the Greenland-Scotland Ridge and towards the Arctic is mainly driven by the SCAN.

In paper V, we examine whether the interplay between the NAO and the

34

other leading modes (SCAN and EAP) is a good indicator of anomalous vol- ume transports towards the Arctic and Barents Sea. In particular, we focus on the flow of Atlantic water associated with the Svinøy branch, the Fram Strait branch and the Barents Sea branch. The reasoning is that a pure NAO phase would only exist if the SCAN and EAP were zero, and hence [Moore et al., 2012] the observed atmospheric circulation in the North Atlantic is composed of these three leading modes (or teleconnections). On monthly time scales, we find that the combination of the positive (negative) phases of NAO and SCAN is the pattern driving the anomalously high (low) transports of all three branches This is particularly prominent for the Fram Strait branch. For the Bar- ents Sea branch, we find that a large majority of the anomalously-low-transport months can be attributed to the positive EAP phase (irrespective of the NAO phase). During winter, NAO is the leading mode controlling the variability of the Svinøy and Fram Strait branches. However, SCAN appears to be the mode of greatest importance for the Barents Sea branch transports. Our find- ings suggest that it is essential to take into account all three leading modes for explaining the anomalous transports of warm Atlantic water towards the Arctic region; the NAO alone does not suffice as a forcing mechanism.

4.2 The Atlantic Multidecadal Oscillation

The variations of the averaged sea surface temperature (SST) in the North At- lantic with a period of 50-70 years [Kushnir, 1994] are often referred to as the Atlantic Multidecadal Oscillation or AMO (Fig. 4.3). This low-frequency SST variability is of major socio-economic importance since it relates to many phenomena of great climatic importance [Knight et al., 2006], e.g. shifts in the inter-tropical convergence zone (Sutton and Hudson, 2005), droughts in the Sahel [Rowell et al., 1995], and the frequency of Atlantic hurricanes [Gold- enberg et al., 2001]. The AMO has also been shown to be closely linked to the variability of the meridional overturning circulation, both in paleo proxy records [Gray et al., 2004] and in climate models [Delworth and Mann, 2000;

Zhang and Delworth, 2006]. Latif et al. [2004] found that anomalies in the sea-surface temperature reflect changes in the strength of the thermohaline cir- culation, but with the caveat that their results were based on a climate model.

Zhang [2008] found, in a 1000 year model simulation, a significant correlation between AMO and the AMOC.

The AMO has also been linked to the atmospheric circulation and gyre dy-

namics. Most recently, Häkkinen et al. [2011b] demonstrated that the AMO

varies in phase with the atmospheric blocking frequency in the northern North

Atlantic. They argued that blocking events weaken the circulation of the gyres

35

Figure 4.3: The Atlantic Multidecadal Oscillation (AMO). Figure courtesy of UCAR.

and reduce the heat loss to the atmosphere, processes that lead to warmer tem- peratures in the North Atlantic and contribute to a positive AMO phase. In paper I, we observe a decreasing trend in the transport between Scotland and Greenland, which most likely is associated with the present positive phase of the AMO (Fig. 4.3), but certainly not with the NAO.

36

5. Outlook

During the last years, European winters were unprecedented in severity and length. The British media, e.g., compared the severity to that in 1816, which was a year with no summer (The Guardian, 2013-March-28). The harsh 2012- 2013 winter (cf. Fig. 5.1) was also observed in Russia and North America, but climate and weather-forecast models failed to predict this event.

Figure 5.1: The map displays surface-temperature anomalies March 14-20, 2013 compared to conditions during the same periods from 2005 to 2012 (red = posi- tive, blue = negative). It is based on data from the Moderate Resolution Imaging Spectroradiometer (MODIS) on the NASA Aqua satellite. Figure courtesy of NASA.

The present thesis has dealt with a considerable range of processes and phe- nomena in a broad swath of the ocean ranging from the northern North Atlantic to the Arctic. Many of the features which have been discussed are well worthy of further investigation, but future research efforts will primarily be directed towards the Barents Sea. This region, denoted BS, is of considerable impor- tance for Arctic climate and European weather variability. The sea-ice cover is directly influenced by the inflowing warm Atlantic waters [Årthun et al., 2012]. A decrease of sea-ice in the BS has been found to i) amplify surface air temperatures over the Arctic [Bengtsson et al., 2004; Koenigk et al., 2013], ii) trigger more frequent cold European winters [Petoukhov and Semenov, 2010;

Yang and Christensen, 2012], and iii) lead to wintertime cooling trends over

37

Eurasia [Outten et al., 2013].

Atmospheric

circulation

Sea ice

Atlantic inflow to the

Barents Sea

winter &

summer conditions

Figure 5.2: Schematic representation of the Barents Sea coupling and feedback mechanisms.

In paper V, we examined the atmospheric circulation patterns leading to anomalous oceanic transports into the BS. In particular, we highlighted that the interplay between the NAO and the other two leading modes of atmospheric variability is a good indicator of the BS inflow variability. To extend this study, we aim at i) linking these combined patterns directly to the sea-ice concentra- tion/extent and ii) investigating the dynamical feedback of the sea-ice in the Barents Sea on the atmospheric circulation, and hence the European weather (during winter as well as summer) and climate (decadal to multidecadal time scales). These processes are schematized in Fig. 5.2. To accomplish this, both observations and CMIP5 (Coupled Model Intercomparison Project Phase 5) model results will be used. Coupled model simulations in the CMIP5 have been shown to have some problems in representing sea-ice extent in both the Arctic and Antarctica [Stroeve et al., 2012; Turner et al., 2012], indicating that there are coupled processes in play that we do not fully understand.

38

Sammanfattning

Föreliggande avhandling behandlar dynamiken och cirkulationen i norra Nor- datlanten och de nordiska haven, processer som är av avgörande vikt för Skan- dinaviens och Nordeuropas milda klimat. Högupplösta ADCP-mätningar av strömfältet ner till ett djup av 400 m mellan Skottland och Grönland visar att Rejkjanesryggen på ett effektivt sätt separerar flödena till de Nordiska ha- ven och Labradorhavet. Det framgick att den meridionala transporten har för- svagats med ca 1.7 Sv (1 Sv = 10 ⁶ m ³ s ⁻¹ ) under den period på 18 år när satellit-altimetriska data varit tillgängliga. Denna trend torde vara en effekt av den atlantiska multidecennieoscillationen, och är helt säkert inte en konse- kvens av den nordatlantiska oscillationen (NAO). Genom att studera cirkula- tionen i Färö-Shetlandsundet, som utgör en viktig stryppunkt för den globala termohalina cirkulationen, kunde slutsatsen dras att kontraktionen av gyren i Norska Havet under perioder med lågt NAO spelar en viktig roll vad gäller strömningsmönstrets fördelning. Specifikt påverkas det regionala havsklima- tet genom att varmt och högsalint atlantvatten ansamlas i sundet. Under pe- rioder med högt NAO kontrolleras cirkulationen av topografin. Den norska atlantsluttningsströmmen (NwASC) är den huvudsakliga strömningsgren som länkar Nordatlanten till Arktis och Barents Hav. Vi fann att att NwASC till största delen är koherent över säsongs- och mellanårliga tidsskalor. För kor- tare tidsskalor uppvisade strömningskoherensen en fortlöpande och påtaglig försvagning nedströms Lofoten. Intensiv virvelavlösning från kontinentalslutt- ningen in i Lofotenbassängen dämpar strömningskoherensen. Virvlarna tar två månader på sig att förflytta sig till och förena sig med den nästan permanenta anticyklonala virvel som är belägen ovan Lofotenbassängens djupaste område.

Dessa resultat har implikationer för hur strömnings- och hydrografiska anoma-

lier överföres genom det Nordiska Havet mot Arktis. Avvikande transporter

av varmt vatten till Arktis och Barents Hav drivs av en kombination av NAO

och de två övriga ledande moderna av den nordatlantiska atmosfäriska vari-

abiliteten. De i avhandlingen redovisade resultaten torde vara av vikt för att

i klimatmodeller tillämpa en korrekt representation av de mot Arktis riktade

värmetransporterna.

Acknowledgements

I would like to acknowledge everyone who has assisted me and helped me in one way or another throughout my university studies over the years.

First of all, I would like to thank Peter Lundberg. Without you, I would not be here today. Thanks for hiring me and for believing in me throughout my PhD. I am for- ever grateful to you. Johan Nilsson, thanks for supervising, inspiring and guiding me all the way. Your immense knowledge in ocean-atmosphere dynamics is impressive.

Your brilliant ideas, suggestions and ways to tackle problems, all kind of problems, is something I will carry with me all my life. I cannot think of a better supervisor to have. Peter Sigray, thanks for all the help with the statistics. Thanks for always asking how things are, both on an academic and personal level.

I started at MISU as an undergraduate and here I am now almost 6.5 years later.

There are many people that inspired me during both my undergraduate and graduate studies. Foremost among these are my first supervisors Erland Källén and Linus Mag- nusson. Thanks for switching on my dynamical thinking and for all your support dur- ing the masters. Also I would like to thank my former teachers Jörg Gumbel, Annica Ekman (who got me interested in synoptics), Michael Tjernström, Jonas Nycander, Gunilla Svensson, and last but not least Kristofer Döös. Heiner Körnich, thanks for all the informal discussions and for all the help before and after I enrolled the PhD program. A very special thanks is given to Rune Grand Graversen for all the inspiring conversations. Frida Bender, thanks for being a great PhD committee member.

The administrative staff (old and new) are highly appreciated. Thanks to Pat Jo- hansson and Janne for all your assistance. Cecilia Törnqvist, thanks for all the help lately with all the paper work. A thousand thanks is specially given to Susanne Eriks- son.

MISU is a really friendly and enjoyable place to work at, and this is thanks to all the great PhD students. Thanks to my old class mates Marcus Löfverström, Jenny Lindvall and Abubakr Salih for always asking how things are progressing. Jonas Mortin and Marie Kapsch, it’s always great talking to you and thanks for all the chats.

A special thanks goes to Raza Ranjha and Saeed Falahat for always making my day at MISU, which included laughs, personal support and all kind of scientif assistance. My time at MISU was also enriched by the French group who deserve a special thanks.

Laurent Brodeau, thank you for all the support at the beginning my PhD and thanks

for teaching me how to program properly. Thanks to Maxime Ballarotta, for all the

chats about life and science. Mondheur Zarroug, thanks for being a good friend since

my undergraduates. I also truly thank Fabien Roquet for introducing me to the South- ern Ocean, and for always taking time discussing with me.

I would also like to thank all my co-authors for their cooperation. The american group: Ryan Sriver and Matt Huber. The Norwegian group and friends: Jan Even Øie Nilsen, Tor Eldevik and Roshin Pappukutty. A special thanks is dedicated to Abdel Hannachi for all your support at the end of my PhD.

I would like to give a special acknowledgment to Joakim Kjellsson. Thanks for being the best ”cellmate” ever. Thanks for your constant support and for always being there when I needed you. Your hard work has truly inspired me during the last 5/11 years. I wish you all the best in England, throughout your career and life.

I owe my deepest gratitude to my inspiration source, my collaborator and my dear friend Thomas Rossby. This PhD would not have been possible without your endless support. Thanks for letting me and my family (including a baby girl) stay at your house. We greatly appreciate your warm hospitality. We will always remember our visit to Rhode Island. I also would like to give a very special thanks to Øystein Sk- agseth for always being there since the beginning of my PhD, for always answering my long list of questions, for always offering thorough and excellent feedback, and for always being a good host in Bergen.

I also wish to thank the Climate Research School and the Bolin Centre for Climate Research for all the travel grants I have received during my PhD. I truly appreciate it!

These last words of acknowledgment I saved to my wife, my best friend and my

love Natasha as well as to my daughter Léa. Thanks for being supportive since day

1 at the university, and making all these years the best of my life. This thesis seemed

like a distant dream, but your encouragement, understanding, faith is what made it

possible.