Detecting Frontal Structures in Oceanic Station
Time Series
J. Even Ø. Nilsen∗‡ and Frank Nilsen†
∗
Geophysical Institute, University of Bergen, Norway
†
UNIS, Norway
(Manuscript)
Abstract
In this paper the geographical spread of hydrographic measurements
at the Ocean Weather Station M (OWSM, 66◦ N 2◦ E) is utilized to create mean sections across the Vøring Plateau Escarpment. These sections
form the basis for a study of the characteristics of the mean spatial hydrography near OWSM. It is shown that the halocline and thermocline at
this position are sloped from about 200 m in the west and down to 400 m
in the east over 40 km centered on the station. The horizontal gradients
introduced by this slope are 2◦ C and 0.1 for salinity, over 40 km. Seasonal
sections clearly reveal a broader and stronger salt core during summer.
Composite sections for high and low winter NAO-index show that there is
a wider (more westward) spread of Atlantic Waters at this position during
and somewhat after periods of high NAO-index. The frontal slope is not
seen to change its inclination between seasons nor decades, indicating that
the dynamic control of this frontal slope does not change appreciably on
these timescales. Further supported by a theoretical framework it is thus
shown that the subsurface part of the Arctic Norwegian Front and the
associated western branch of the Norwegian Atlantic Current is strongly
locked by topography along the Vøring Plateau.
1
Introduction
By definition, fronts are regions of large horizontal gradients of certain properties, usually including density. Fronts are marked by large thermocline/halocline
slopes, large in the sense that the depth variation of an isotherm/isohaline across
‡ Corresponding author. Tel: +47 55588697; fax: +47 55589883.
Email: [email protected] (J.E.Ø. Nilsen)
1
VøringPlateau
NorwegianBasin
Iceland
Z
Z
Z
QQ
Q
Z
Z
Norway
Q
Faroes
Q
Q
Q
Figure 1: The Norwegian Basin and adjacent areas. Whole arrows indicate the two
current branches of the Norwegian Atlantic Current: An eastern barotropic flow following the Norwegian Continental Slope, and a western baroclinic jet following the
Atlantic Norwegian Front as a continuation of the Faroe Current along the Iceland
Faroe Front. The fronts are indicated by the dot-dashed line. Question marks indicate uncertainty as to areas of substantial interchange of water masses, and uncertainty
as to the persistence of the baroclinic current over the weaker bottom relief. Straight
lines indicate the Svinøy (northernmost) and Sognefjord Section (southernmost).
a front is typically on the order of the isotherm/isohaline depth itself. In the
Nordic Seas the collision of cold, fresh waters of the Arctic and the warm, saline
waters of the North Atlantic produces such large thermocline and halocline
slopes (which again produces large pycnocline slopes and dynamical effects),
and thus the large horizontal gradients in what is called the Arctic Front (AF).
The strength of this front is primarily determined by the properties of the water
masses, i.e. the horizontal density gradient produced by the two water masses.
The topography and Atlantic surface currents in the Southern Norwegian Sea
are shown in Figure 1. Ocean Weather Ship Station M (OWSM, 66◦ N 2◦ E) is
situated over the steep slope from the Vøring Plateau down to the Norwegian
Basin. The current system of the Norwegian Atlantic Current (NWAC) is considered to be a two branch system with a topographically controlled barotropic
current along the Norwegian Continental Slope and a baroclinic jet following
2–2500 m isobaths (Mork and Blindheim, 2000; Orvik et al., 2001; Orvik and
2
Niiler, 2002). This jet is connected to the subsurface part of the southeastern part of the AF, the Atlantic Norwegian Front (ANF; Szczechowoski, 1994;
Smart, 1984). At the the Svinøy-section, the western branch is identified as a
400 m deep and 30–50 km wide frontal jet (Orvik et al., 2001).
The distribution of Atlantic Water (AW) is spread over the whole area between these two current branches, and to a certain extent outside but then at
shallower depths. This spreading is due to extensive eddy activity (Sælen, 1963;
Rodionov, 1992) and atmospheric forcing. The surface distribution varies considerably with season and other atmospheric influences (Blindheim et al., 2000;
Furevik et al., 2002; Nilsen and Falck, 2003). The NWAC is often characterized
as a wedge shaped “current”. However, already from the early measurements of
Helland-Hansen and Nansen (1909) it was clear that the lower boundary of the
AW (35.0 isohaline) spreads in an uneven fashion, but horizontally on average,
as far as to the 2000 m isobath before showing any persistent upward sloping
(see Figure 2). The severe vertical displacements of this boundary, first referred
to as “puzzling waves” by Helland-Hansen and Nansen (1909), are elaborated
upon thoroughly by Dickson (1972). They are most likely explained by the existence of large eddies in the inflowing AW (Sælen, 1963), although the alternative
explanations, such as internal waves (Krauß, 1958) and “sudden variations in
the velocity ... of the surface currents” (Helland-Hansen and Nansen, 1909),
can be viewed as different formulations of the same dynamics.
The upward westward sloping of isotherms and isohalines at 2000 m water depth is characteristic of most cross sections of the NWAC (e.g. Figure 2).
North of the Faroes, inflowing AW enters the Norwegian sea as the Faroe Current along the continental shelf break. Here the AW forms the clearly defined
sloping Iceland-Faroe Front against the colder and fresher waters of the Icelandic
and western Norwegian Sea (Hansen and Østerhus, 2000). The Sognefjord Section cuts across the bottom of the ridge from the north-east corner of the Faroe
Plateau, and here the front is clearly visible NW of the ridge (Helland-Hansen,
1934; Sælen, 1959). Over the weak slope further north, at the Svinøy section,
the mentioned unstable baroclinic frontal jet is located approximately above the
2000 m isobath (Leinebø, 1969; Mork and Blindheim, 2000; Orvik et al., 2001).
Such a slope is also seen in the 6S section along 65◦ 45’N (Borovkov and Krysov,
1995; Anon., 1997; Blindheim et al., 2000) wherefrom the three years 1996–1998
are shown in Figure 3. Hydrographic surveys of the area around and south of
the Vøring Plateau in 1935, ’36 and ’65 (Mosby, 1959, 1970; Bjørgen, 1971),
show highly variable hydrography, both in time and space, and the existence of
eddies. But the outer slope of the AW interface and calculated geostrophic surface current maxima most often coincide with the position of OWSM. Across the
Vøring Plateau at the latitude of OWSM (66◦ N) mostly sporadic measurements
are done, the most focused on this area being the Anglo-Norwegian Variability
Project during seven weeks in the spring of 1967 (Kvinge et al., 1968). Dickson
3
Figure 2: Salinity section across the Norwegian Sea between Norway and Iceland from
a R/V Håkon Mosby cruise in August 1994. From Østrem (1998).
(1972) studied the waviness of the isolines in these 17 sections and a few others, and found that the frontal slope at OWSM was the most pronounced and
persistent feature here. Other sections along 66◦ N are shown in Gammelsrød
and Holm (1984), and the importance and possible implications of frontal movements on the time series from OWSM are pointed out.
Another type of spatial study in this area are AXBT (Airborne Expendable
Bathythermograph) surveys from January 1980 to July 1981 (Smart, 1984).
This dense dataset covering all seasons, reveals the ANF at around 2–400 m
depth, following the 2000 m isobath along the Vøring Plateau Escarpment with
a lateral variation in position as small as about 50 km. He also observed that the
magnitude of the frontal slope did not appear to be seasonally dependent, but
the slopes did show regional dependence on the local steepness of the bottom
topography. All these surveys indicate a sloping interface between AW and
Arctic Intermediate Water, above the same isobaths as the ones below OWSM.
That the surface current in the western branch of the NWAC follows the
frontal structure is shown by drifters (Poulain et al., 1996; Orvik and Niiler,
2001). These drifters show the swiftest surface currents to follow the steepest
slopes and the clearest separation of the two branches of the NWAC to be across
the Vøring Plateau. Using a two layer model Heburn and Johnson (1995) show
a clear dependence on topography for the circulation, but they do not resolve
4
a)
b)
o
o
Temperature June 1996 along 65.75 N
0
54
3
−0.5
343
.948.
96
34.92
−500
0
−1000
34.9
1
34.91
−1000
−1500
Depth [m]
Depth [m]
3592
34.
6
2
1
−500
Salinity June 1996 along 65.75 N
0
8
7
−2000
−2500
−1500
−2000
−2500
−3000
−3000
65.75oN
66oN
−3500
−4
−3
−2
−1
0
1
2
Longitude
3
4
5
65.75oN
66oN
−3500
6
c)
−4
−3
−2
−1
0
1
2
Longitude
3
4
5
6
d)
Temperature June 1997 along 65.75oN
0
Salinity June 1997 along 65.75oN
0
7
2
1
3
0
34.91
−500
34.92
34.98 35
34.96
−1000
−1000
−1500
−1500
Depth [m]
Depth [m]
34.9
1
−500
1
6
4
3
35.2
34.9
5
−2000
−2500
−2000
−2500
−3000
−3000
o
o
65.75 N
66oN
−3500
−4
−3
−2
−1
0
1
2
Longitude
3
4
5
65.75 N
66oN
−3500
6
e)
−4
−3
−2
−1
0
1
2
Longitude
3
4
5
6
f)
Temperature June 1998 along 65.75oN
Salinity June 1998 along 65.75oN
0
0
2
1
35.2
34.8
7
34.91
5
−500
4
6
3
34
.98
−500
−1000
−1000
−1500
−1500
Depth [m]
Depth [m]
34.9
1
0
−2000
−2500
−2000
−2500
−3000
−3000
65.75oN
66oN
−3500
−4
−3
−2
−1
0
1
2
Longitude
3
4
5
65.75oN
66oN
−3500
6
−4
−3
−2
−1
0
1
2
Longitude
3
4
5
6
Figure 3: Temperature (left) and salinity (right) in the Russian standard section along
65◦ 45’N (6S; Borovkov and Krysov, 1995) which has been worked every June since
1963. The hydrographic section data from June ’96 (a,b), ’97 (c,d) and ’98 (e,f) was
provided by PINRO, Murmansk.
5
the two branch structure of the NWAC very well. Using particle tracking in the
Princeton Ocean Model (POM) Hjøllo (1999) clearly demonstrate a tendency
for most of the Atlantic Water in the Faroe Current to follow the 2000 m isobath
towards and along the Vøring Plateau slope.
Theoretical grounds for guidance by topography of a baroclinic current in
the upper layer is shown by Svendsen et al. (1991), using a two layer analytical
model. This control by the relation
v · ∇H = 0,
(1)
where v is upper layer velocity and H total water depth, requires the relative
vorticity in both layers to be negligible in the long term. As mentioned, the area
around the ANF is influenced by eddies shed from the front and meanders of the
front itself (Rodionov, 1992). However, the current field is observed to reestablish after perturbations and thus have a quasistationary character (Kort et al.,
1977). Thus the long term current pattern will obey the topographic control.
A thorough study of the dynamics of perturbations and locking mechanisms of
the front over the Vøring Plateau slope is given by Nilsen (2001).
In this paper the geographical spread of hydrographic measurements at the
OWSM will be utilized to create mean sections across the Vøring Plateau Escarpment, to see if the sloping front of the NWAC can be identified using these
time series. Sections created for the four seasons, and sections for high and low
NAO-index (composite) will also be presented. These sections form the basis
for a study of the characteristics of the mean spatial hydrography near OWSM,
with focus on the sloping pycnocline. The extent to which this subsurface front
is locked to the bottom slope from the Vøring Plateau will be examined from
the sections, and a physical explanation for the mechanisms behind the topographical locking on shorter timescales will be given.
2
Materials and Methods
Figure 4 shows the distribution of all the measurements taken at OWSM between 1948 and 1999. The measurements of temperature and salinity are taken
around the standard depths of 0, 10, 25, 50, 75, 100, 150, 200, 300, 400, 500, 600,
800, 1000, 1200, 1500 and 2000 m, using Nansen bottles equipped with reversing
thermometers. The program carried out at the station consists of daily casts
down to 1000 m, and weekly down to the bottom (Gammelsrød et al., 1992).
Counting every single sample (not profiles), the dataset consists of some 94 000
values of each variable. As can be seen, the precision in positioning has not always been perfect, due to the necessary priority of the synoptic meteorological
measurements. Note also that the spread of measurements roughly covers the
whole escarpment. After the creation of a Cartesian coordinate system with
origin at 66◦ N 2◦ E and rotated 17◦ anticlockwise (Figure 5a), it is possible to
6
a)
b)
Figure 4: The data samples in space, horizontal (a) and vertical (b) distribution. Gray
lines represent isobaths. The axes in (b) are rotated for an along slope perspective.
study the spatial distribution of the samples (Figure 5b,c). The system’s alignment is based on the assumption that a topographically steered jet will follow
the escarpment, and that the associated sloping front can be found studying a
section perpendicular to this. The exact choice of rotation angle is based on
minimization of the gradients along the y-axis, and this makes the coincidence
with the isobaths (Figure 5a) a clear indication that the front is aligned with the
escarpment. The samples are almost normally distributed, and the bulk of measurements are found inside a ±20 km interval in both directions (Figure 5b,c).
The variation in time of the deviation from 66◦ N 2◦ E can be seen in Figure 6.
The largest spread is found in the beginning of the series and during the ’70s.
For calculation of an overall mean section, the cross-slope section was divided
into nine 10 km wide bins laterally and 17 bins centered on the standard depths
(Figure 7). All samples in time and y-direction falling inside a bin form the
basis for the mean value for this bin. The section along the escarpment (ydirection) was created in the same manner.
The temporal inhomogeneity of the sample spread (Figure 6) has implications on this spatial study, in that any values calculated for far away locations
will be based on data only from these specific periods in time. The number of
profiles in the outer bins (at 40 km) are around 70 increasing to more than 1500
at 10 km and 5000 at the center (Figure 5b). Their total count being relatively
low, the outer bins reflect the overall temporal distribution with most samples
during the ’50s and ’70s. Strictly speaking, from this the only bins with appreciable homogeneity in time are in the 5 central columns (± 20 km, see Figure 6).
The annual distribution of samples on the other hand, is homogeneous in all
7
a)
b)
c)
68 oN
x−distribution
y−distribution
6000
6000
5000
5000
4000
4000
66 oN
Profiles
Profiles
67 oN
3000
3000
2000
2000
1000
1000
65 oN
0
64 oN o
2 W
0o
o
4 E
2oE
−50 −40 −30 −20 −10
o
6 E
0 10 20 30 40 50
x [km]
0
−50 −40 −30 −20 −10
0 10 20 30 40 50
y [km]
Figure 5: a) The Cartesian coordinate-system used in this study. The y-axis is aligned
along the bottom slope, with positive direction northward. The length of the arrows
is approximately 50 km. Panels b) and c) show distribution of measurements along
the two axes.
bins. The error estimate for the mean values (sm ) is calculated following
s2m =
s2
,
n
(2)
where s is the estimated standard deviation, and n the number of samples in
the bin.
In addition to the overall mean sections, differences between the four seasons
and between high and low NAO-phases are studied by cross-slope sections of
bin-means from the same dataset. Due to the reduction of sample number in
this temporal separation, and the scarcity of data far from 66◦ N 2◦ E, the width
of the sections had to be reduced to 40 and 20 km respectively. But given the
large amount of samples within these distances of the station position, it was
possible to increase the resolution of the sections to bin sizes of 4 km and 2.5 km.
Distance from 2°E 66°N
150
r [km]
100
50
0
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
Figure 6: Time series of deviation from designated position.
8
6
∼2 km
?
¾
∼90 km
-
Figure 7: The bins in which mean values are calculated. Nine 10 km wide bins laterally
(x-axis) and 17 bins centered on the standard depths (0–2000 m, y-axis).
3
3.1
Results and Discussion
Overall Mean Sections
The resulting cross-stream mean salinity section is shown in Figure 8a. The
section clearly shows a halocline sloping from about 200 m in the west and
down to 400 m in the east. There is also a fresher surface layer, which is related
to the fresher shallow summer mixed layer in this area (Nilsen and Falck, 2003).
The mean sections of temperature show the same characteristics with warmer
water above and to the east, and a sloping thermocline at the same depths as
the halocline (Figure 9a). The frontal slope gives mean horizontal gradients
of 0.1 in salinity and 2◦ C over 40 km at these depths. The frontal structure is
present in both salinity and temperature, so it can be expected to be for density
also. Still, since the density distribution determines the dynamics of a system,
a density section is therefore shown in Figure 10.
The along slope (y) sections (Figures 8b–10b) show a distinctly more flat
structure, with small or nonexistent horizontal gradients. There is a slight
upward tilt in isotherms and isohalines toward south, and the possible dynamical
explanation for this lies in the eastward bend in the isobaths south of OWSM
(Figure 5a) which a topographically controlled front will tend to follow. The
contrasts between the two perpendicular sections leave no doubt that the mean
situation involves a frontal structure aligned with the escarpment.
Also embedded in the plots (Figures 8–10) are error estimates given by (2)
in white contours. These show the most unreliable mean values to be near
9
b)
5
0.005
1
0.0
06
0
0.0
0.004
4
34.95
35.05
0.0
02
0.002
35
0.002
05
0.0
600
S
0.01
0.004
0.01
06
34.95
2
0.0
0.01
0.005
0.005
0.01
6
0.00
500
0.008
0.00
0.0
d [m]
35.1
0.008
06
4
35.15
0.008
0.00
400
08
0.0
0.01
0.
08
0.0
5
0.0
300
0.00
01
6
35.2
1
0.00800.0
.01
35.15
0.008
0.0
1
0.00
8
06
0.004
0.00
0.0
0.005
0.008
200
01
4
00
35.15
0.006
0.
0.
05
0.0
0.006
100
0.018
0.016
0.014
0.012
0.01
0.01
0.006
001
.0
0.0.
168
01
4 00.008
0.012
0.0
0.01 .01
0.00
065
08
00.0
.01
8
00
0.
0
0.0
0
a)
34.95
700
800
−40
34.9
−20
0
x [km]
20
40
−40
−20
0
y [km]
20
40
Figure 8: The mean (1948–99) section of salinity across (a) and along (b) the bottom
slope. Black contour lines indicates variance in the bins. Whole white contour lines
indicate error estimate for the bin mean values, with contours chosen at 1/10 and 2/10
of the contour steps for the water property. Dashed white lines indicate the σt =27.8
isopycnal found from mean sections of density anomaly (Figure 10).
10
b)
3
2
0.2
3
2
4
0.
1
5
0.2
3
2 0.1
4
0
0.2
a)
1
9
1
100
1
1
7◦ C
0.1
7◦ C
1
40.2
3
20
.
4
1
1
0.1
5
0.2
o
2
0.2
400
2
2
1
1
1
1◦ C
3
0.
1◦ C
0.1
500
0.2
1
0.1
d [m]
2
3
2
2
300
7
0.1
2
T [ C]
0.2
0.1
0.1
1
200
1
0.1
600
1
700
800
−40
−1
−20
0
x [km]
20
40
−40
−20
0
y [km]
20
40
Figure 9: The mean (1948–99) temperature section. Details as in Figure 8.
11
a)
b)
0
28.2
0.01
0.008
0.01
0.008
27.5
08
0.
0.0
0.0
11
28
0
0.006
0.0
0.0
00
6
1
08
1
0.01
0.00
8
27.8
0.01
0.01
0.01
2
0.0
0.0
0.0
0.008
.01
0.006
200
0.0
400
27.6
08
0.01
0.0
1
0.0
08
0.0
06
0.008
0.006
0.00
4
0.01
06
1
0.0
6
.00
0.010 04
0
.
0
28
4
0.00
27.4
0.01
0.0
500
0.008
σt
300
d [m]
2
27.5
0.01
0.02
0.01
100
0.01
0.0
0.02
0.01
0.0
04
28
27.2
600
27
700
800
−40
26.8
−20
0
x [km]
20
40
−40
−20
0
y [km]
20
40
Figure 10: The mean (1948–99) density section. Details as in Figure 8.
12
the edges of the sections as expected due to the scarceness of data here. But
within 20 km of the middle position, the errors are less than 1/10 of the contour
intervals used to plot the property. This means that the patterns seen are
reliable. The variance within the bins (i.e. the distribution of variance over the
section) is shown with black contours.
3.2
Seasonality
The two branches of the NWAC react differently to the changes of seasons. In
the Faroe Current, the saline AW is more widely distributed in summer than
in winter (Hansen et al., 2000), and Mork and Blindheim (2000) found that
the core of AW furthest offshore at the Svinøy-section is less distinct in the
winter and spring. To investigate the horizontal characteristics and the frontal
behaviour across the slope over the year, the data from OWSM was divided into
four seasons using the same months as Hansen et al. (2000), and mean sections
for the three water properties were created.
The salinity sections clearly reveal a broader, somewhat deeper, and more
saline core in the summer (Figure 11b and c).
This can be due to the
variability in AW flow in the western branch of the NWAC. There is also an
apparent horizontal shift in the Atlantic water masses, but the fresher AW
column in the early winter is not caused by such a lateral shift in the NWAC.
Instead the salinity cycle is caused by vertical mixing in autumn of the fresh
summer layer seen in Figure 11b and c. This layer originates mostly from
the Norwegian Coastal Current and enters as Ekman transports in the shallow
summer mixed layer (Nilsen and Falck, 2003).
The temperature sections in Figure 12 do not show any large horizontal
differences between the seasons, and this is due to the atmospheric heat fluxes’
dominant influence on the temperature in the upper waters. The density section
(Figure 13) shows the strong stability of the warm and fresh summer mixed layer
near the surface, but since the density variations in the upper water masses at
OWSM are almost entirely determined by temperature variations, there is no
sign of the lateral structure seen in the corresponding salinity sections.
The more interesting result from these seasonal sections with respect to the
question of frontal locking, is that the frontal slope (slope of the halo-, thermoand pycnocline) is similar through all seasons. That is, not only is the slope
roughly constant, which alone would not tell us anything about lateral shifts,
but the depth of the front in the middle of the section is also the same. Note
that this does not exclude variability in frontal position on other timescales than
the annual, but given the generally strong energy in the annual cycle, this is an
important result.
13
Feb − Apr
a)
0.004
500
0.00
4
5
01
0.
0.006
300
d [m]
08
0.004
0.002
0.008
0.01
0.002
5
00
500
0.
600
0.006
0.00
2
0.0
0.010
5
Aug − Oct
8
0.014
1
0.016
0.00.001
0.012
8 0.0
0.008
8
00
0.0
200
0.0
02
0.0
0.00
6
0.004
0.0
050.002 0.01
800
−20
5
0.00
0.00
4
0.0
06
0.008
0.008
0.006
0.004
0.002
0
x [km]
10
20
800
−20
35
5
0
.0
34.95
0
34.9
−10
0
x [km]
10
20
Figure 11: Cross slope salinity distribution in four seasons. Dashed white lines indicate
the σt =27.8 isopycnal found from seasonal sections of density anomaly (Figure 13).
Other details as in Figure 8.
3.3
Relation to the NAO-index
Mork and Blindheim (2000) found that the temperature and salinity in the western part of the Svinøy Section are negatively correlated with the NAO-index.
Studies of the distribution of AW in the surface layers across the Norwegian
Sea also show a more wide (westward) spread of these waters some 2–3 years
after periods of low NAO winter index, and a narrower surface signature after
high index periods, attributed to changes in the pathways into the Nordic Seas
(Blindheim et al., 2000). Supporting this, a model study by Nilsen et al. (2003)
shows a long term anti-correlation between the inflow on each side of the Faroes.
This correlation is related to the NAO so that during high index years the eastern branch of the NWAC receives more AW from the North Atlantic than the
western branch.
14
35.15
35.05
700
−10
35.2
35.1
0.004
0.000.0
21
600
0.
00
5
700
0.00
0.016
0.01
500
0.0
05
400
0.0
05
0.006
0.01
300
08
0.006
0.0
04
600
100
0.01
1
0.
400
006
05
0.00.
20
1
0
0.008 0.
0.01
10
Nov − Jan
0
0.004
200
0
x [km]
0.0
05
0.01
100
−10
d)
00.02
.01 0.014 00.0
.011
028
0.0
64
0.00
0.00
d [m]
0
800
−20
20
S
10
0.004
0
x [km]
5
−10
c)
d [m]
0.006 0.008
0.004
5
0.002
00
0.
700
800
−20
500
0.01
6
0.00
600
700
300
400
0.0.0
01
0.0
612
14
0.0
0.006
0.004
0.
00
d [m]
400
0.0
200
05
1
0.0
0.006
0.01
0.0086
0
0.0
0.005
300
100
0.02
0.016
0.01
0.014
0.012
0.01
0.
0.0581
0.00
6 80.01
00
0.00
0.006
0.
0.
00
01
0
4
0.01 .008
0.0
0.008
200
0
0
0.004
0.01
0.006
100
May − Jul
b)
5
.00
0.
00
0
Feb − Apr
a)
0 3
1
500
12
0.
600
d [m]
2
2
0.1
0.1
1 0.2
10
0.1
400
500
600
0.1
1
2
2
0.1
800
−20
−10
0
x [km]
10
5
1
1
3
0.1
0.1
600
700
7
0.2
2
d [m]
0.2
1
500
9
300
2
0.1
2
0.1
200
2
1
20
100
0.1 1
300
10
0.1
1
2
0
x [km]
Nov − Jan
0
1
0.2
0.1
0.2
−10
d)
02.1
10.1
800
−20
20
1
700
800
−20
20
−1
−10
0
x [km]
10
20
Figure 12: Cross slope temperature distribution in four seasons. Details as in Figure 11.
The OWSM-data from winters (Dec–Apr) in longer periods of relatively
high (’73–’76, ’81–’84, ’89–’95) and low (i.e. weak or negative, ’68–’72, ’77–’80,
’85–’88) winter NAO-index (Figure 14) were separated to make corresponding
mean-sections. The results in Figure 15 show that also at OWSM there are
changes related to these periods in time. The salinity sections (Figure 15a) show
the same relation to the NAO-index as cited above, with a fresher Atlantic layer
in high index years than in low, at this location. In the high index situation,
the most saline waters are found in the eastern part of the section, i.e. there is
a stronger lateral gradient in the upper waters. This horizontal shift in water
mass characteristics in the AW seems to tilt the isohalines down to the east
of OWSM and consequently strengthen the horizontal gradient in the frontal
slope as predicted by Nilsen (2001). This also corresponds to the results of
15
T [ oC]
0
x [km]
Aug − Oct
2 0.2
1
1
200
d [m]
2
10.
0.2
−10
c)
400
0.1
0.1
10.2
700
800
−20
100
0.22
2
600
700
0
400
500
1
0.1
0.1
1
0.1
300
2
2
1
200 1
1
2
0.2
d [m]
0.1
0.1
0.1
2
0.1
400
100
0.1
1
300
0.2
2
4
100
200
May − Jul
b)
0.1
0
200
0.006
00.008
.01
0.01
0.008
0.0 0.0.006
1 004
400
500
600
600
700
700
−10
0
x [km]
10
Aug − Oct
0.02
4
00
100
0.0
200
0.00
2.01
700
10
01
.
0.008 0
0.006
0.004
0.01
800
−20
20
0.01
0.01
0.008
0.006
28.2
0.0081
0.0
27.8
28
27.6
27.4
0.0
0.0008
0.0064
500
700
0
x [km]
20
1
400
600
−10
6
0.00
300
600
800
−20
1
0.01
0.0
0.01
0.00
0.0 6
04
10
Nov − Jan
8
00
08
500
0.0
0.008
0.0
0.0 06
04
400
0.006
0.004
0.01
0
x [km]
0.
0.006
0.008
300
0.02
0.01
0.01
0.
0.00.01 01
2
200
0.
100
−10
0
0.02
0.01
0.01
0.006 0.008
0.00
6
0.008
0.006
0.004
0.01
d)
04
0
0.0
c)
0.004
0.01
0.01
800
−20
20
d [m]
800
−20
0.01
0.01
0.004
0.008
300
0.02
0.008
0.006
0.0104
0
0.
500
d [m]
0.01
0.02
0.02
400
4
0.0
1
0.008
0.006
0.01
0.006
σt
300
0.00
100
0.02
0.02
0.01
0.006
0.008
0.01
0.01
200
d [m]
0
0.01
100
May − Jul
b)
0.004
d [m]
0.01
0.0
1
0
0.02
Feb − Apr
a)
27.2
27
26.8
−10
0
x [km]
10
20
Figure 13: Cross slope density distribution in four seasons. Details as in Figure 11.
Blindheim et al. (2000). In the periods where the surface front in salinity is
found close to the eastern side of the Norwegian Basin, sections from OWSM
show the same eastward shift, and vice versa for the periods with widely spread
AW in the basin. Lagging our periods by the suggested 2 years, yields even
clearer differences in salinity, supporting the findings of Blindheim et al. (2000).
It is hard to discuss these results in terms of NAO-impact, since the chosen
periods are long and prone to be influenced by long term variability. Occurrence
of events like the Great Salinity Anomalies (Belkin et al., 1998; Dickson et al.,
1988; Haak et al., 2003) and the fact that the ’60s was a period of high salinity
both at OWSM and in the return flow on the western side of the basin (Alekseev
et al., 2001), influence the interpretation of the results here as well as those of
Blindheim et al. (2000). Independent of their causes, it is interesting to set
the two clearly different situations in contrast to each other, to see if these
16
3
Winter NAO−index
2
1
0
−1
−2
−3
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
Figure 14: The North Atlantic Oscillation (NAO) winter index (December to March
mean) as defined by Jones et al. (1997). In the winter, the difference between the
normalized sea level pressure over Gibraltar and the normalized sea level pressure over
Southwest Iceland is a useful index of the strength of the prevailing westerly winds in
the North Atlantic.
changes in water mass distribution has any impact on the frontal structure (i.e.
density gradients). In the same manner as between the seasons, the temperature
(Figure 15b) does not show any significant changes between high and low NAO
periods, and thus the calculated density structure is not significantly affected
either (not shown). This will be further addressed in section 3.4.
3.4
A Locked Front
The literature reviewed in the introduction clearly indicate both from observations and theory that the baroclinic outer branch of the NWAC and its associated subsurface front are guided by topography. The question remains however,
whether this is true for all timescales, and especially to what extent the ANF
is locked above the steep Vøring Plateau Escarpment. From the sections in
Figures 8–13 and 15, this can only be qualitatively assessed. Apart from the
changes in water mass distribution in the upper waters, all the cross sections
seem to have a similar slope in the different properties.
The most important of these properties for localization of the front and the
position of the baroclinic jet is of course density. In Figures 8–13 and 15, the
isopycnal of σt =27.8 is plotted as representative of the frontal slope. In addition
to the cases already shown, a separation between periods of basin wide and
limited spread of AW from the NWAC according to Blindheim et al. (2000) was
done. As expected, this showed basically the same salinity distributions as for
the NAO-study in Figure 15a, only with an even clearer difference (Section 3.3).
All these isopycnals are gathered in Figure 16a.
The slopes show a frontal
structure basically indifferent to the changes in upper layer water properties
inferred by the annual cycle and the interannual to decadal variability in phase
17
a)
0.006
0.00. 04
01
2
00
0.
0.005
5
0. 0
0.004
02
0.0
00
5
0.00
35.1
0.006
35.05
0.004
0.002
01
0.
1
600
0.01
0.006
0.0
500
00
0.
0.
700
35.15
35
5
0.004
400 0.002
0.01
0.0
0
1008
300 0.00 0.0.
60
004
00.
.0
06
35.2
S
0.00
0.06
08
1
0.0
0.0.01
0
0.006 08
0.01
0.01 04
0.0
1
0.0
4
0.
0.008
200
d [m]
0.00
100 0.006 01
Low Winter−NAO
0.004
0.005
0.01
04
High Winter−NAO
0
5
34.95
34.9
800
−10
0
x [km]
10−10
0
x [km]
10
b)
3
400
2
0.2
2 2
2
1
0.2
0.1
500 1
0.1
1
0.1
1
0.2
9
0.2
1
7
2
5
1
0.2
0.1
0.1
3
1
0.1
700
800
−10
0.21
0.1
0.1
0.1
0.1
02.2
d [m]
2
1
0.2
300
600
0.1
0.1
100
200
Low Winter−NAO
0.1
o
0.1
T [ C]
High Winter−NAO
0
−1
0
x [km]
10−10
0
x [km]
10
Figure 15: The mean cross slope sections of salinity (a) and temperature (b) from
high and low NAO-index-winters respectively. Dashed white lines indicate the σt =27.8
isopycnal found from corresponding sections of density anomaly (not shown). Other
details as in Figure 8.
18
a)
b)
260
260
Overall mean
Feb − Apr
Mai − Jul
Aug − Oct
Nov − Jan
NAO+
NAO−
Eastward SSS
Westward SSS
280
300
Overall mean
’50s
’60s
’70s
’80s
’90s
280
300
340
d [m]
320
340
d [m]
320
360
360
380
380
400
400
420
420
440
−40
−30
−20
−10
0
x [km]
10
20
30
440
−40
40
−30
−20
−10
0
x [km]
10
20
30
Figure 16: Cross slope depth of the σt =27.8 isopycnal representing the pycnocline at
OWSM, for different situations (composite studies) (a) and for the different decades
(b).
with the NAO. The steepest part of the slope deepens from 280 to 400 m depth
over 40 km centered on OWSM. The lateral spread of curves from the different
situations are only on the order of 4–8 km in the range of appreciable amounts of
data (±10 km). More importantly, the spread consists of random irregularities
in each slope rather than a discernible separation of slopes. Any such separation
or differences in steepness of slope fall inside this range of noise. The details in
the seasonal slopes outside the 10 km range can not be given any significance due
to the scarceness of data there. Thus if the part of the frontal slope encompassed
by these sections is representable or at least part of the front, difference in the
current speed in the western branch of the NWAC between high and low NAO
periods or in an annual cycle is not likely.
As mentioned in Section 3.3, the composite study of periods of high and low
NAO-index is likely to be influenced by long term changes in hydrography not
necessarily (at least not directly) related to the NAO. Thus the interpretation
here must instead be in terms of high and low salinity rather than the NAOindex itself. In connection to this it is interesting to note that the time series
from OWSM shows long term variability in depth of isopycnals (Nilsen, 2003c),
and to study this, mean sections from the five last decades have been produced
(not shown) and the pycnocline slopes plotted in Figure 16b. The depth of these
isopycnals show a sloping isopycnal at varying depths: In the ’50s approximately
20 m shallower, in the ’70s deeper by almost the same amount, while in the other
decades close to the mean. The steepness of the slope does not change due to
the long term changes. Note that the use of the term “depth” in relation to
the slopes in Figure 16 is somewhat inappropriate since no leveling is seen at
the ends of the slope in the composite study and a “deepening” might translate
19
40
as a westward shift. It is not possible from these data to separate between the
two, although it will be argued that the frontal slope has less degrees of freedom
laterally due to the topography, than it has vertically where changing water
mass characteristics may alter the structure of the water column.
The short term variability is hard to isolate from composite studies of these
data. As already shown, separating the dataset into subsets strongly limits
the lateral extent of spatial studies from station time series. The fact that the
sloping front has a mean position as shown in the different sections, does not
exclude the possibility for lateral excursions. Horizontal gradients in the mean
sections will be somewhat smoothed by lateral movements and other changes in
the front, but a mean gradient will still be present.
Near a front, variability is connected to movement in the strong gradients
of the front. A front that “spends” just as much time away from the position
of OWSM as in the middle of it, will leave a broad area of variability. Thus
the frontal slope can be localized by maxima in variance, and in Figures 8–
13 and 15 the variance within the bins (i.e. the distribution of variance over
the section) is shown with black contours. All these sections have variance
maxima focused around the sloping front. The presence of a variance maximum
around the front is not unexpected, but their slanted character and narrowness
relative to the seen gradient of the property does indicate that frontal excursions
from the mean position is limited. This localisation of a frontal structure and
variability maxima is also shown by Dickson (1972, Figure 136) in his study of
17 consecutive sections along 66◦ N. The maximum range of T and S between
these sections is found not only to be focused along the pycnocline (as expected),
but also at two lateral maxima, namely along the sloping part of the pycnocline
(under OWSM) and at the interception with the continental slope. The latter
is merely the result of a stronger gradient there, while the former is clearly
connected to the existence of a persistent isopycnal slope at this position.
4
4.1
Dynamical Explanation for Topographic Front
Trapping
The Barotropic Assumption
Rhines (1970) considered the propagation of planetary waves in a stratified
fluid with a rigid lid and a uniformly sloping bottom. By nondimensionalizing
the governing equation and boundary conditions, one of the nondimensional
combinations of the chosen scales was the Burger number,
³
B(z) =
N (z)D
f0
´2
L2
20
µ
=
Ri (z)
L
¶2
,
(3)
where N is the Brunt-Väisälä frequency, D is the characteristic depth, f0 is the
local Coriolis parameter, and Ri is the internal Rossby radius of deformation.
The Burger number is thus the square of the ratio of the internal Rossby radius of
deformation to the horizontal length scale (L). Willebrand et al. (1980) found
that a forcing function with scales much larger than O(100 km) and periods
between the inertial period and ∼ 300 days, induced oceanic motion that was
depth independent (barotropic response), the baroclinic part being a direct local
response to the Ekman pumping. The oceanic response to large scale forcing
can differ strikingly from this if the bottom topography is taken into account.
Then the horizontal length scale is determined by the slope width and the scales
of forcing and ocean response are no longer directly related. As discussed by
Willebrand et al. (1980), the answer to the question whether or not the flow will
be baroclinic, now depends on the topographic as well as the atmospheric scale.
If the atmospheric scale is assumed to be much larger than that of topography,
the wind induced barotropic flow produces a uniform vertical velocity field at the
bottom, which has the same scales as the topography. This vertical velocity field
can induce either an additional barotropic signal or a baroclinic motion trapped
near the bottom. By using the Burger number (3) it can be determined which of
the two is occurring. Rhines (1970) found that the nondimensional parameter
relevant for determining the vertical structure of the horizontal velocity was
given by
p
R = Ri k 2 + l 2 ,
(4)
where k and l are the horizontal wave numbers for the planetary waves. Hence,
the wave length becomes the important horizontal scale. Motions associated
with topographic scales larger than the internal Rossby radius, affect the whole
water column while motions with smaller horizontal scales remain trapped near
the bottom.
The barotropic response of the ocean is described by the vertically integrated
equations of motion. From these equations the quasigeostrophic potential vorticity (QPV) equation can be derived. Care must be taken in the top and
bottom of the water column since wind stress and bottom friction produces
vertical velocity through Ekman pumping/suction. The vortex tube stretching by the Ekman pumping velocity acting on the planetary vorticity filaments
will produce a variation of the relative vorticity. This is a nongeostrophic effect arising from the influence of friction. Although the water column is highly
baroclinic at OWSM and along the slope from the Vøring Plateau, investigation of topographically trapped waves can be done by means of a barotropic or
homogeneous model including friction. The underlying assumption is that the
internal Rossby radius of deformation is much smaller than the characteristic
horizontal wavelength. Thus the effect of topography during the spin-up phase
is felt mainly by the barotropic mode. The dominant frequencies which will
be observed during the spin-up of a baroclinic ocean are almost exactly those
21
of the purely barotropic modes (Rhines, 1970; Anderson and Killworth, 1977;
Nilsen, 2001, 2003a,b). The focus on the spin-up phase is due to the fact that
the spatial wind stress field over the Norwegian Sea is highly fluctuating: As
soon as a low pressure center has started to get a hold of the water column,
it either moves away from the area and free oscillations are released, or it is
replaced by a new pressure center and the spin-up starts again.
Measurements at OWSM (Figure 13) and from the Russian section along
65◦ 45’N in the Norwegian Sea (6S; Figure 3; Borovkov and Krysov, 1995; Anon.,
1997; Blindheim et al., 2000) show that the water column can be approximately
represented by a two layer model with an upper layer thickness between 200 and
400 m in the winter and spring. The Rossby deformation radius for a two-layer
model with an upper layer H1 and a lower layer H2 is given by
Ri =
ci
,
f0
c2i = g
ρ 2 − ρ 1 H1 H2
H1 H2
= g0
,
ρ 2 H1 + H2
H1 + H2
(5)
where ci is the phase speed for the baroclinic wave, and g 0 = g ∆ρ
ρ2 is the reduced gravity. Using the mid-depth of the sloping bottom as the characteristic water depth, i.e. D = H0 = 2250 m, H1 = 400 m, ρ1 = 1027.6 kg m−3 ,
ρ2 = 1028.1 kg m−3 and f0 = 1.33·10−4 s−1 , ci =1.3 m s−1 and Ri =9.7 km. Thus
for a horizontal scale of 73 km, which is the characteristic width of the Vøring
Plateau Escarpment (Nilsen, 2001, 2003b), stratification will play a marginal
role since B ∼ 0.02 when (5) is used in (3). But as shown by Rhines (1970)
in (4), the horizontal wavelength of the planetary wave becomes the important
horizontal parameter when determining the vertical structure of horizontal velocity. One may expect topographic effects to decay with the Prandtl e-folding
scale (Barnier and Le Provost, 1993)
he =
fL
,
N
(6)
and in Rhines’ (1970) work he is the e-folding scale for the vertical structure
of linear waves on a sloping bottom. For the Vøring Plateau slope R =O(1),
and, using a realistic N found in the mean pycnocline from OWSM profiles,
he ∼ 4850 m for wave lengths associated with topographic scales (L=73 km).
Hence, the topographic effects should be felt through the water column and the
stratification could influence the motions on these scales. Based on the results
in Nilsen (2003a), where it is shown that the forcing by the variability of the
pressure systems over the Vøring Plateau can be represented by a wave length
on the order of the Vøring Plateau Escarpment length, the barotropic mode is
assumed to be the dominant mode in the spin-up phase.
22
4.2
Forced Waves and Spin-up
Here we will characterize the forced water column and spin-up phase through
the QPV-equation. A rigid lid approximation is used and the effect of viscosity
on the interior is neglected. The lateral diffusion of vorticity is also neglected by
assuming the Reynolds number Re → ∞ in the interior. The topography dominates over the β-effect (Nilsen, 2003b), and thus a constant Coriolis parameter
is used. By these approximations and linearisations, the QPV-equation for a
barotropic response of the ocean to a wind stress is described by
µ
¶
∂
∇ψ
Hx ∂ψ
1
H
∇·
+f
= k · (∇ × τ ) = C(x, y, t),
(7)
∂t
H
H ∂y
ρr
where H = H(x) is the Vøring Plateau slope, Hx = dH/dx, τ is the wind stress
vector, and ψ is the volume transport stream function defined so that
Hu = −
∂ψ
,
∂y
Hv =
∂ψ
.
∂x
(8)
1
Using the transformation ψ = H 2 ϕ in (7) yields
¡ 2
¢ ∂ϕ
Hx ∂ϕ
C
∇ − V (x)
+f
= 1,
∂t
H ∂y
H2
where
1
1
V (x) = H 2 ∇2 H − 2 .
(9)
(10)
V (x) is identified as the potential function of the topography, described more in
detail by Nilsen (2003b). Equation (9) can be viewed as an advection equation
for the wave form ϕ forced by the expression on the right hand side. The wave
form is advected along the y-direction,
¡ i.e. along¢ the escarpment. In absence of
the forcing term, the fraction f HHx / ∇2 − V (x) can be interpreted as a phase
speed for the topographically trapped wave propagating in the y-direction.
An important fact from (9) is that, for waves over topography, f HHx takes
the role of β for planetary waves. The magnitude of this term determines how
apparent these waves will be in the area of interest, and also what horizontal
length scale these waves will have. Plotting the reciprocal of HHx , based on the
ETOPO5 bathymetry for the Nordic Seas, reveal where there most likely will
exist topographically trapped waves and what their horizontal scales will be
(Figure 17).
23
o
80 N
-1
50
0
00 0
-10 -200
110
o
78 N
-4
00
-3000
90
-1
00
-30
-4 000
00
-1-1050
00
-30
000
o
50
0
-2
76 N
-3000
0 00
--41-010500
0
[km]
00
H
Hx
0
-40
0
00
-1 -2000
70
-2
o
74 N
50
-3000
-400
-20
72 N
30
-2000
-1500
0
-150
-150
0
00
-15
-2000
0
70 N
10
00
-400
o
-1
-200 000
0
00
-30
-1
20
-1
00
0
-41000000
- -15
o
50
0
-1
00
-1
0
50
-4
00
0
-100
-3
00
-30
0
-400
-400
0
0
-100
00
-1000
-1
5
-4 00
00
00
-20
00
-20
-400
00
-1 -15
00 00
0
000
-15 -1
00
-1
5
-1
o
00 00
-15 -10
0
0
0
62oN
00
0
0
-10
o
20 E
-15
-200
-4
0
o
16 E
-1
0
00
00
0
-1
00
-1
0
-40 0
-100 00
0
-2
-20
00
-200
00
-15
o
64 N
-400
0
0
-4
00
-4
0
-150
0 -1000
-40
-2000
-2000
0
50
-1
-400
o
66 N
-1000
-3000
00
-10
-400
0
00
00
-4
-40
-30
0
00
-1
-1500
68 N
0
-1000
o
-400
0
50
-3000
60 N
o
40 W
o
36 W
o
32 W
o
28 W
o
24 W
o
o
20 W 18 W
o
14 W
o
10 W
o
6 W
o
o
2 W 0
o
o
2 E 4 E
o
8 E
o
12 E
Figure 17: Contours of HHx values based on the ETOPO5 bathymetry data for the
Nordic Seas, showing where there most likely will exist topographically trapped waves
and what their horizontal scales will be.
24
o
24 E
4.3
Front Trapping
As shown by Nilsen (2003a) the wind stress curl over the Vøring Plateau slope
can contain high level energy on the period interval between 48 and 185 hours.
The forcing term in (9) can be represented by a forcing function constructed as
a sum of (infinitely) many Fourier components, and Nilsen reported that if this
forcing function contained variability corresponding to a wave length of ∼ 76 km
and an oscillating period of ∼ 100 hours, a resonant response for topographically
trapped waves can occur over the slope. These are also the characteristic lengthand time scales for the Vøring Plateau slope around OWSM (Nilsen, 2003b).
The calculated wind stress curl field in spring ’97 showed a larger variability
over the Nordic Seas compared with ’96 and ’98 (Nilsen, 2003a). Although the
cross-slope CTD sections in Figure 3 are only snapshots in time, they show that
the front was more defined and sharper in ’97 compared to the other two years.
Combining the atmospheric forcing information with the hydrographical data
through the theory presented above, could give an explanation on the frontal
structure over the sloping sea floor: The larger variability in the wind stress
curl is a result of strong atmospheric pressure systems traversing the Nordic
Seas. If the wind stress curl field contains variability corresponding to the
length- and time scales supported by the Vøring Plateau slope, the response will
primarily be a barotropic perturbation of the water column. Hence, information
of a sloping ocean floor is communicated to the dynamics in the whole water
column. Pressure systems over the Vøring Plateau are soon replaced by new
strong pressure systems, and thus the spin-up process starts over again with a
possible barotropic response. Although these perturbations can initiate current
meandering and eddy formations they also serve to trap the frontal- and current
system above the slope (Nilsen, 2003a). Therefore, a more defined and sharp
front can be found in years with large time variability in the wind stress curl
field.
As shown by Svendsen et al. (1991), using a two layer analytical model,
a baroclinic current in the upper layer can be guided by topography. Topographical steering requires a velocity in the lower layer in order to communicate
information about the sloping bottom to the upper layer. A recent publication
by Nøst and Isachsen (2003) shows that there is a northward geostrophic bottom current along the Vøring Plateau slope, hence the theoretical arguments
in Svendsen et al. (1991) are sufficient to explain the long term topographical
steering. The forcing mechanism discussed in this paper is thus an extension
of the mean picture, in an attempt to explain the monthly to annual variation
in the sloping front and also the air-ocean interaction processes responsible for
trapping the front along the Vøring Plateau slope.
25
∼50 km
-
Figure 18: Schematic of the sloping pycnocline at OWSM. Several types of perturbations can give rise to high variability in time series at constant depth near such
a gradient (from left to right): An increase in differences between the neighbouring
water masses will both change the strength of the gradient and alter the tilt of the
slope; the lateral position of the slope may exhibit short term shifts; frontal waves and
instabilities give water movements perpendicular to the slope; passing internal waves
and eddies introduce additional perturbations to the structure, and also important
are the changes in the ship’s position. The dashed vertical lines indicate the observed
scale of ship deviation relative to the width of the frontal slope.
5
A Note on the Influence on Time Series
It has been shown that there is a sloping of the pycnocline more or less locked
above the bottom slope at OWSM’s position. The existence of this frontal
structure and its associated surface jet has implications for the time series from
the station. A simple conceptual model of this issue is given in Figure 18. A
level pycnocline by itself introduces higher variability in constant depth time
series sampled near its location in the water column, since any perturbation of
the interface will place the samples in different water masses in a more or less
arbitrary manner. A sloping pycnocline and the pertaining baroclinic jet have
dynamics of their own, adding instabilities/meandering and eddy shedding to
the sources of short term variability. The frontal structure can also change its
inclination, strength of gradient and lateral position due to changes in properties
and volume of the interfacing water masses. And even without variability in the
front itself, its horizontal gradients introduce sensibility to ship positioning.
This effect on the measurements at OWSM was early recognized by Mosby
(1950). He showed a variation of the depth of the transition layer with as much
as 300 m in chosen profiles taken within a week of each other. Using the spatial
survey of “Armauer Hansen” in 1936, Mosby points out as the reason for this
that OWSM is positioned in the middle of a frontal structure. Later studies of
26
OWSM-data also hold forth perturbations of this sloping frontal structure as
an explanation for rapid changes in the hydrographical records (Le Floch, 1953;
Mosby, 1959; Johannessen and Gade, 1984; Gammelsrød and Holm, 1984).
6
Concluding Remarks
In this paper we have identified the sloping Atlantic/Arctic front over the
Vøring Plateau Escarpment by utilizing the spread of measurements from Ocean
Weather Station M. The use of profiles from a single station to create spatial
sections of hydrography is shown to be applicable for producing long term mean
sections as well as composite studies. There are severe limitations connected to
decreasing amounts of data with distance from the stations’ designated position.
The sections from OWSM clearly show a halocline and thermocline sloping
from about 200 m in the west and down to 400 m in the east over the 40 km
covered, with warmer and more saline waters above and to the east, resulting
in considerable horizontal gradients.
Seasonal sections reveal a broader and stronger salt-core in the summer, supporting the findings of more saline and widely distributed AW in the western
branch of the NWAC (Hansen et al., 2000; Mork and Blindheim, 2000). Composite sections from periods with high and low winter NAO-index show that at
OWSM there is a more wide (westward) spread of Atlantic Waters during periods of high NAO-index, coherent with the findings of Blindheim et al. (2000)
and Mork and Blindheim (2000).
A more central result is that the frontal slope, represented by the midpycnocline σt =27.8 isopycnal, is shown to be indifferent to changes in upper
layer hydrography inferred by the annual cycle and the interannual variability
in phase with the NAO. Furthermore, although the vertical position of the
slope varies on decadal timescales, no change is seen in its steepness. This
indicates that although the hydrographic conditions in the water column and the
neighbouring waters change, the dynamical control of this frontal slope and its
associated jet (the western branch of the NWAC) does not change appreciably
on timescales from annual to decadal. Together with the narrowness of the
variance associated with the frontal gradient, this implies that the subsurface
part of the Arctic Norwegian Front is strongly locked by topography along the
Vøring Plateau.
A physical explanation of a topographically locked baroclinic current and
its short term behaviour is presented through the quasigeostrophic potential
vorticity equation. It is argued that the air–ocean interaction through the wind
stress curl field is responsible for trapping the front along the Vøring Plateau
slope. When the variability of the wind stress curl field contains the length- and
time scales supported by the slope, the response will primarily be a barotropic
perturbation of the water column. Hence, information of a sloping ocean floor
27
is communicated to the dynamics of the whole water column. Although these
perturbations can initiate current meandering and eddy formations they also
serve to trap the frontal- and current system above the slope. Therefore, on
monthly to annual time scales, a more defined and sharp front is found in years
with strong temporal variability in the wind stress curl field.
Outlook
In light of the current monitoring techniques developed for the eastern branch
in the Svinøy Section (Orvik and Skagseth, 2002), it would be interesting to
assess the possibilities for a current meter program over the steeper Vøring
Plateau Escarpment. In general, further field studies on the stability of frontal
positions and behaviour over escarpments are needed. Recently a promising
method is developed where temperature gradients and thus oceanic fronts can
be identified and studied from seismic reflection profiling (Holbrook et al., 2003),
giving possibility for the use of the vast amounts of data already collected in
seismic surveys of the past. Furthermore, as alluded to in Figure 1 and in light
of Figure 17, the amount of exchanges of AW between the two branches needs
to be assessed in order to monitor the fluxes in the whole NWAC with a higher
degree of accuracy.
This paper also show that caution should be taken when interpreting time
series from OWSM, especially from earlier decades. Nilsen (2003c) showed that
the effects of vertical, or rather, cross isopycnal movements in the stratification
(pycnocline) at OWSM can be eliminated, and more or less purely advective
signals can be studied by interpolating the measurements from standard depths
onto surfaces of constant density. However, when vertical movements are of interest the time series are usually studied at constant depths, and then the measurements’ relative position to horizontal gradients becomes important (both in
terms of movement of the fronts and the measuring position). Further effort
should be made to assess these effects on the time series and the possibility to
correct for varying lateral position of the ship.
Acknowledgments
Many thanks to Svein Østerhus for helping with the OWSM data. Vladimir Borovkov
and PINRO (Murmansk) are acknowledged for providing the CTD section along
65◦ 45’N.
28
References
Alekseev, G. V., Johannessen, O. M., Korablev, A. A., Ivanov, V. V., and Kovalevski,
D. V. (2001). Interannual variability of water mass in the Greenland Sea and the
adjacent areas. Polar Research, 20(2), 201–208.
Anderson, D. L. T. and Killworth, P. D. (1977). Spin-up of a stratified ocean, with
topography. Deep-Sea Res., 24, 709–732.
Anon. (1997). Tables with Averages of Temperature, Salinity and Normal Values in
Standard Sections in the Norwegian and Greenland Seas (1983-1995). Technical
report, Knipovich Polar Research Institute of Marine Fisheries and Oceanography
(PINRO), Murmansk. In russian.
Barnier, B. and Le Provost, C. (1993). Influence of bottom topography roughness
on the jet and inertial recirculation of a mid-latitude gyre. Dyn. Atmos. Oceans,
18(1–2), 29–65, doi:10.1016/0377-0265(93)90003-P.
Belkin, I. M., Levitus, S., Antonov, J., and Malmberg, S.-A. (1998). “Great Salinity
Anomalies” in the North Atlantic. Prog. Oceanog., 41(1), 1–68.
Bjørgen, H. (1971). “Armauer Hansen” ekspedisjonene i Norskehavet 1935 og 1936.
Master’s thesis, Geophysical Institute, University of Bergen, Allégt. 70, 5007 Bergen,
Norway.
Blindheim, J., Borovkov, V., Hansen, B., Malmberg, S. A., Turrell, W. R., and Østerhus, S. (2000). Upper Layer Cooling and Freshening in the Norwegian Seas in
Relation to Atmospheric Forcing. Deep-Sea Res., 47, 655–680.
Borovkov, V. and Krysov, A. (1995). Report from servey of distributions of Norwegian
spring spawning herring and environmental conditions in the Norwegian Sea in JuneJuly 1995. In Report from a Joint Meeting of Russian and Icelandic Scientists in
Reykjavik 1995.
Dickson, R. (1972). Variability and continuity within the Atlantic Current of the
Norwegian Sea. Rapp. P.-v Réun. Cons. int. Explor. Mer, 162, 167–183.
Dickson, R. R., Meincke, J., Malmberg, S.-A., and Lee, A. J. (1988). The “Great
Salinity Anomaly” in the northern North Atlantic 1968–1982. Prog. Oceanog., 20,
103–151.
Furevik, T., Bentsen, M., Drange, H., Johannessen, J. A., and Korablev, A. (2002).
Temporal and spatial variability of the sea surface salinity in the Nordic Seas. J. Geophys. Res., 107(C12), 8009, doi:10.1029/2001JC001118.
Gammelsrød, T. and Holm, A. (1984). Variations of temperature and saltinity at
Station M (66◦ N 02◦ E) since 1948. Rapp. P.-v Réun. Cons. int. Explor. Mer, 185,
188–200.
29
Gammelsrød, T., Østerhus, S., and Godøy, Ø. (1992). Decadal variations of ocean
climate in the Norwegian Sea observed at Ocean Station “Mike” (66◦ N 2◦ E). In
R. Dickson, P. Mälkki, R. S. G. Radach, and M. Sissenwine, editors, Hydrobiological
Variability in the ICES Area, 1980-89, volume 195 of ICES Mar. Sci. Symp. ICES.
Haak, H., Jungclaus, J. H., Mikolajewicz, U., and Latif, M. (2003). Formation
and propagation of great salinity anomalies. Geophys. Res. Lett., 30(9), 1473,
doi:10.1029/2003GL017065.
Hansen, B., Jónsson, S., Turrell, W. R., and Østerhus, S. (2000). Seasonal Variations
in the Atlantic Water inflow to the Nordic Seas. ICES CM 2000/L, (3), 1–15.
Hansen, B. and Østerhus, S. (2000). North Atlantic–Nordic Seas Exchanges. Prog.
Oceanog., 45, 109–208.
Heburn, G. W. and Johnson, C. D. (1995). Simulations of the mesoscale circulation
of the Greenland-Iceland-Norwegian Seas. J. Geophys. Res., 100(C3), 4921–4941.
Helland-Hansen, B. (1934). The Sognefjord Section. oceanographic observations in the
northernmost part of the North Sea and the southern part of the Norwegian Sea.
In James Johnstone Memorial Volume. University Press of Liverpool, UK, pages
257–274.
Helland-Hansen, B. and Nansen, F. (1909). The Norwegian Sea: Its physical oceanography. In J. Hjort, editor, Rep. Norw. Fish. Mar. Inv., volume II. The Royal Department of Trade, Navigation and Industries, Oslo, Norway, pages 1–390.
Hjøllo, S. S. (1999). A Comparative Study of the Norwegian Sea Inflow. Ph.D. thesis,
Geophysical Institute, University of Bergen, Allégt. 70, 5007 Bergen, Norway.
Holbrook, W. S., Páramo, P., Pearse, S., and Schmitt, R. W. (2003). Thermohaline
fine structure in an oceanographic front from seismic reflection profiling. Science,
301, 821–824.
Johannessen, J. A. and Gade, H. G. (1984). A case study of the variations in the upper
ocean at Ocean Weather Ship Mike (66◦ N, 2◦ E) in the Norwegian Sea. Geophys.
Norv., 32(5), 165–175.
Jones, P. D., Jonsson, T., and Wheeler, D. (1997). Extension to the North Atlantic
Oscillation using early instrumental pressure observations from Gibraltar and SouthWest Iceland. Int. J. Climatol., 17, 1433–1450.
Kort, V., Titov, V., and Osadchiy, A. (1977). Kinematics and structure of currents in
a study area in the Norwegian Sea. Oceanology, English Translation, 17(5), 505–508.
Krauß, W. (1958). Our knowledge about internal waves in the northern North Atlantic
Ocean. ICES CM 1958, (54), 3. (mimeo).
Kvinge, T., Lee, A., and Sætre, R. (1968). Report on Study of Variability in the
Norwegian Sea April/May 1967. Report 15, Geophysical Institute, University of
Bergen, Allégt. 70, 5007 Bergen, Norway.
30
Le Floch, J. (1953). Quelques propriétés de la zone de transition entre eau Arctique
en Mer de Norvège. Univ. Bergen Årbok, Mat.-Naturv. Rekke, (3), 45.
Leinebø, R. (1969). Den Norske Atlanterhavsstrøm. Rapport fra målinger høsten 1962.
Cruise report, Geophysical Institute, University of Bergen.
Mork, K. A. and Blindheim, J. (2000). Variation in the Atlantic Inflow to the
Nordic Seas, 1955–1996. Deep-Sea Res. I, 47(6), 1035–1057, doi:10.1016/S09670637(99)00091-6.
Mosby, H. (1950). Recherches océanographiques dans la mer de Norwège a la Station
Météorologique M. Cahiers du Centre de Recherches et D’etudes Océanographiques,
1, 1–7.
Mosby, H. (1959). Deep water in the Norwegian Sea. Geophys. Norv., 21(3), 1–62.
Mosby, H. (1970). Atlantic Water in the Norwegian Sea. Geophys. Norv., 28(1), 1–60.
Nilsen, F. (2001). On the Effect of Atmospheric Forcing and Topography at the Vøring
Plateau. Ph.D. thesis, Geophysical Institute, University of Bergen, Allégt. 70, 5007
Bergen, Norway.
Nilsen, F. (2003a). Forcing of a two-layered water column over a sloping sea floor.
J. Phys. Oceanogr. Submitted.
Nilsen, F. (2003b). Topographically trapped waves along the Vøring Plateau slope Theory. J. Mar. Res. Submitted.
Nilsen, J. E. Ø. (2003c). Variability at Ocean Weather Station M in the Norwegian
Sea. In W. Turrell, A. Lavı́n, K. F. Drinkwater, M. St. John, and J. Watson, editors,
Hydrobiological Variability in the ICES Area, 1990-99, volume 219 of ICES Mar.
Sci. Symp. ICES.
Nilsen, J. E. Ø. and Falck, E. (2003). Variations of mixed layer depth and water properties in the Norwegian Sea during 50 years. In W. Berger, T. Dokken, H. Drange,
T. Furevik, and R. Gerdes, editors, Climate Variability in the Nordic Seas. Geophysical Monograph Series, AGU. Submitted.
Nilsen, J. E. Ø., Gao, Y., Drange, H., Furevik, T., and Bentsen, M. (2003). Simulated North Atlantic–Nordic Seas water mass exchanges in an isopycnic coordinate
OGCM. Geophys. Res. Lett., 30(10), 1536, doi:10.1029/2002GL016597.
Nøst, O. A. and Isachsen, P. E. (2003). The large-scale time-mean ocean circulation
in the Nordic Seas and Arctic Ocean estimated from simplified dynamics. J. Mar.
Res., 61(2), 175–210, doi:10.1357/002224003322005069.
Orvik, K. A. and Niiler, P. (2002). Major pathways of Atlantic water in the northern
North Atlantic and Nordic Seas towards arctic. Geophys. Res. Lett., 29(19), 1896,
doi:10.1029/2002GL015002.
31
Orvik, K. A. and Niiler, P. P. (2001). Circulation of atlantic water in the northern
North Atlantic and Nordic Seas. ICES CM 2001/W.
Orvik, K. A. and Skagseth, Ø. (2002). Monitoring the Norwegian Atlantic Slope
Current using a single moored current meter. Cont. Shelf Res., 23(2), 159–176,
doi:10.1016/S0278-4343(02)00172-3.
Orvik, K. A., Skagseth, Ø., and Mork, M. (2001). Atlantic Inflow to the Nordic
Seas. Current structure and volume fluxes from moored current meters, VM–
ADCP and SeaSoar-CTD observations, 1995–1999. Deep-Sea Res. I, 48(4), 937–957,
doi:10.1016/S0967-0637(00)00038-8.
Østrem, A. K. (1998). Det intermediære vannet i Norskehavet. Master’s thesis, Geophysical Institute, University of Bergen.
Poulain, P.-M., Warn-Varnas, A., and Niiler, P. (1996). Near surface circulation of
the Nordic Seas as measured by lagrangian drifters. J. Geophys. Res., 101(C8),
18 237–18 258.
Rhines, P. B. (1970). Edge-, bottom, and Rossby waves in a rotating stratified fluid.
Geophys. Fluid Dynamics, 1, 273–302.
Rodionov, V. (1992). On the mesoscale structure of the frontal zones in the Nordic
Seas. J. Marine. Sys., 3, 127–139.
Sælen, O. H. (1959). Studies in the Norwegian Atlantic Current. part I: The Sognefjord
Section. Geophys. Norv., 20(13), 1–28.
Sælen, O. H. (1963). Studies in the Norwegian Atlantic Current. part II: Investigations
during the years 1954-59 in an area west of Stad. Geophys. Norv., 23(6), 1–82.
Smart, J. H. (1984). Spatial variability of major frontal systems in the North AtlanticNorwegian Sea area: 1980–1981. J. Phys. Oceanogr., 14, 185–192.
Svendsen, E., Sætre, R., and Mork, M. (1991). Features of the northern North Sea
circulation. Cont. Shelf Res., 11(5), 493–508.
Szczechowoski, C. (1994). A robust technique for computing modal paths for oceanic
fronts (abstract). In Eos Trans. AGU, volume 73. Ocean Sciences Meeting suppl.
Willebrand, J., Philander, S., and Pacanowski, R. (1980). The oceanic response to
large-scale atmospheric disturbances. J. Phys. Oceanogr., 10, 411–429.
32