The
Arctic Front (AF) in the Norwegian Sea is an important biologically
productive region which is well-known for its large feeding schools of
pelagic fish. A suite of satellite data, a regional coupled ocean–sea ice
data assimilation system (the TOPAZ reanalysis) and atmospheric reanalysis
data are used to investigate the variability in the lateral and vertical
structure of the AF. A method, known as “singularity analysis”, is applied
on the satellite and reanalysis data for 2-D spatial analysis of the front,
whereas for the vertical structure, a horizontal gradient method is used. We
present new evidence of active air–sea interaction along the AF due to
enhanced momentum mixing near the frontal region. The frontal structure of
the AF is found to be most distinct near the Faroe Current in the south-west
Norwegian Sea and along the Mohn Ridge. Coincidentally, these are the two
locations along the AF where the air–sea interactions are most intense. This
study investigates in particular the frontal structure and its variability
along the Mohn Ridge. The seasonal variability in the strength of the AF is
found to be limited to the surface. The study also provides new insights into
the influence of the three dominant modes of the Norwegian Sea atmospheric
circulation on the AF along the Mohn Ridge. The analyses show a weakened AF
during the negative phase of the North Atlantic Oscillation (NAO
Ocean fronts are boundaries between distinct water masses with large gradients in temperature or salinity (e.g. Bakun, 1996; D'Asaro et al., 2011). The Arctic Front (AF; Swift and Aagaard, 1981; Piechura and Walczowski, 1995) is one of the most prominent ocean fronts in the Norwegian Sea. Similar to its counterparts in the world ocean, the AF is an important biologically productive region also known for its large feeding schools of pelagic fish (e.g. Holst et al., 2004; Blindheim and Rey, 2004; Melle et al., 2004). On its influence on higher trophic levels, it is important to note that Jan Mayen Island located near the AF is an important breeding region inhabited by large colonies of seabirds (Norway Ministry of Environment, 2008–2009).
The Nordic Seas with schematic water pathways showing the northward flowing Atlantic Water in the surface (red) and southward flowing East Greenland Current (black). The two branches of the Norwegian Atlantic Current, the Norwegian Atlantic slope current (NwASC) and Norwegian Atlantic front current (NwAFC) are represented by red arrows. The cyclonic gyre circulations in the Norwegian Basin, Lofoten Basin and Greenland Basin are indicated in blue. See Chatterjee et al. (2018) and Raj et al. (2015) for details. Grey isobaths are drawn for every 600 m. The location of the Arctic Front in the Norwegian Sea coincides with the location of the NwAFC.
The AF in the Norwegian Sea extends from the Iceland–Faroe Plateau to the Mohn–Knipovich Ridge (Nilsen and Nilsen, 2007) and is associated with the interaction of the warm and saline Atlantic Water and the cold and fresher Arctic Water (Blindheim and Ådlandsvik, 1995). As shown in Fig. 1, the Atlantic Water is carried into the Norwegian Sea via the Norwegian Atlantic Current (e.g. Orvik et al., 2001; Raj et al., 2016), which is a two-branch current system, with an eastern branch following the shelf edge as a barotropic slope current, and a western branch following the western rim of the Norwegian Sea as a topographically guided front current (Poulain et al., 1996; Orvik and Niiler, 2002; Skagseth and Orvik, 2002; Orvik and Skagseth, 2003). These two branches are known as the Norwegian Atlantic Slope Current (NwASC; Skagseth and Orvik, 2002) and the Norwegian Atlantic Front Current (NwAFC; Mork and Skagseth, 2010) respectively. On its poleward journey, the NwASC bifurcates, one part flowing into the Barents Sea and the other continuing towards the Fram Strait as the core of the West Spitsbergen Current (Helland-Hansen and Nansen, 1909; Aagaard et al., 1985). The NwAFC on its way to the north encounters three deep currents (Fig. 1); one over the Mohn Ridge flows in the opposite direction (Orvik, 2004). The NwAFC continues poleward, topographically guided by the Mohn Ridge and the Knipovich Ridge, as the western branch of the West Spitsbergen Current (WSC; Walczowski and Piechura, 2006). The Atlantic Water carried by the western branch of the West Spitsbergen Current recirculates mostly within the Nordic Seas, as the Return Atlantic Water (RAW; Eldevik et al., 2009). This AW mixes with the Polar Waters transported south from the Arctic by the East Greenland Current and forms the hydrographically distinct Arctic Water (Blindheim and Østerhus, 2005). The further interaction of the Arctic Water with warm and saline Atlantic Water results in the AF (Swift, 1986). The location of the AF follows the topography and coincides with the location where Arctic Water meets the Atlantic Water.
The AF has been subject of many studies (e.g. Piechura and Walczowski, 1995; Nilsen and Nilsen, 2007). However, the impact of the large-scale atmospheric forcing on the spatial (lateral) and vertical variability of the front has not yet been described. Our study aims to examine the structure (lateral and vertical) of the AF using satellite and reanalysis data, on climatological-mean and seasonal-mean timescales (1991–2015). The near-permanent structure of the AF justifies its investigation over long timescales. In addition, the study aims to examine the influence of the three dominant modes of the large-scale atmospheric forcing in the Norwegian Sea, i.e. the North Atlantic Oscillation (NAO), the East Atlantic Pattern (EAP) and the Scandinavian Pattern (SCAN), on the variability of the AF. The NAO is the most dominant atmospheric mode in the North Atlantic and Nordic Seas (Fig. 2). Even though the impact of the NAO on the NwASC is well-known, its impact on the NwAFC and on the AF is not clearly documented. Furthermore, it has been reported that the location of the centres of the NAO dipole can be affected through the interplay with EAP and SCAN teleconnection patterns (e.g. Moore et al., 2012; Chafik et al., 2017). The independent effect of EAP and SCAN on the AF has also been not investigated yet. We perform a composite analysis (see Sect. 2) on the monthly ocean and atmospheric reanalysis data in order to capture the variability and thus to delineate the impact of the three main atmospheric modes of the Norwegian Sea on the AF. In Sect. 2, we describe the different data sets and methods used in this study. The results are presented in Sect. 3 and are discussed and summarised in Sect. 4.
The first three EOF modes of deseasoned and detrended ERA Interim MSLP (1991–2015) multiplied by the standard deviation of the corresponding principle components. The number in the top right corner of each panel indicates the percentage of variance explained.
TOPAZ is a coupled ocean and sea ice data assimilation system for the North
Atlantic and the Arctic that is based on the Hybrid Coordinate Ocean Model
(HYCOM) and the ensemble Kalman filter data assimilation, the results of its
fourth version (TOPAZ4) have been extensively validated (e.g. Lien et al.,
2016; Xie et al., 2017; Chatterjee et al., 2018). The ocean model
in TOPAZ4 is an eddy-permitting model with 28 hybrid z-isopycnal layers at a
horizontal resolution of 12 to 16 km in the Nordic Seas and the Arctic.
TOPAZ4 represents the Arctic component of the Copernicus Marine Environment
Monitoring Service (CMEMS) and is forced by the European Centre for
Medium-Range Weather Forecasts (ECMWF) Reanalysis (ERA Interim) and
assimilates measurements including along-track altimetry data, sea surface
temperatures, sea ice concentrations and sea ice drift from satellites along
with in situ temperature and salinity profiles. The monthly TOPAZ4 results
used in this study for the time period 1991–2015 have been obtained via
CMEMS (
Advanced Microwave Scanning Radiometer (AMSRE; 2002–2011; 0.25
The monthly detrended and deseasoned ERA Interim MSLP data are used to
calculate the three leading empirical orthogonal functions (EOFs) of
atmospheric variability in the North Atlantic (80
For composite analysis, the positive/negative phase of the NAO
(NAO
There are several methods used in previous studies to identify ocean fronts from satellite data (e.g. Cayula and Cornillon, 1992, 1995; Garcia-Olivares et al., 2007; Turiel et al., 2008). One of them is singularity analysis, which has been established as a powerful tool to detect frontal structures and streamlines from satellite images (Turiel et al., 2008, 2009; Portabella et al., 2012; Lin et al., 2014; Umbert et al., 2015). In 1992, Mallat and Huang introduced singularity analysis of scalar variables in the context of wavelet analysis. Singularity analysis aims to obtain a dimensionless measure known as the singularity exponent at each point, which represents the degree of irregularity at that location. Singularity exponents are dimensionless and can be derived from any scalar quantity, e.g. SST (Turiel et al., 2009) and wind components (Portabella et al., 2012; Lin et al., 2014). A singularity exponent is a continuous extension of classical concepts such as continuity or differentiability. The main difference between the maximum gradient method and singularity exponents is that singularity exponents are normalised so that the absolute value of the gradient is irrelevant. What is important is the degree of correlation between nearby gradients: singularity exponents are the dimensionless measures of that correlation. Hence the results from the singularity analysis of different scalar variables (for, e.g. SST and wind speed) can be directly compared. Furthermore, singularity analysis does not require knowledge of the velocity field, because it is a Eulerian method exploiting the scaling properties of the spatial correlations of the gradients of a given scalar field. Thus, singularity analysis has an advantage over the use of Lyapunov exponents (Garcia-Olivares et al., 2007), another widely used methodology which requires the velocity field to be known.
Our study uses singularity analysis to detect the AF in the Norwegian Sea
from satellite data (SST and wind speed) as well as from ocean reanalysis
data (temperature). Singularity exponents of the above scalar variables have
been estimated using an online service provided by the Barcelona Expert
Center (
Climatological mean (2002–2009)
The spatial distribution of the climatological (2002–2009) SST of the
Norwegian Sea, shown in Fig. 3a, illustrates a typical example of the
distribution of surface waters in the Norwegian Sea. The figure shows that
the warmest waters are confined to the eastern Norwegian Sea, while the
coldest waters are found in the Greenland Basin west of the Mohn Ridge. From
a closer inspection, the comparatively large temperature gradient associated
with the AF at the Mohn Ridge (Piechura and Walczowski, 1995) can be
identified. This large temperature gradient over the Mohn Ridge is associated
with the warm Atlantic Water in the Lofoten Basin and the colder Arctic
Waters in the Greenland Basin. In addition to SST, we use remotely sensed
winds to identify the ocean fronts in the Norwegian Sea, as has been done for
parts of the global ocean (e.g. Song et al., 2006). Figure 3b shows the
spatial distribution of the climatology (2002–2009) of wind speed in the
Norwegian Sea. Note that the choice of time-averaged fields assists in
avoiding possible contamination of the wind field by the rapid evolution of
weather patterns (Chelton et al., 2004). Comparison of Fig. 3a and b reveals
stronger/weak winds over warm/cold waters. Even though shown for the first
time in the Norwegian Sea, the increase/decrease in surface wind speed when
it blows from cold/warm to warm/cold water is well-known, and the physical
mechanism has been extensively studied (e.g. Friehe et al., 1991; Chelton et
al., 2004; Song et al., 2006). The response of the surface wind to the
temperature fronts has been studied on different timescales (monthly, Chelton
et al., 2001; seasonal, O'Neill et al., 2003; and climatological, Chelton et
al., 2004) especially in the western boundary currents. Stronger surface
winds over warmer waters are due to efficient turbulent convection that
transfers momentum down to the surface (Wallace et al., 1989). This process
known as the momentum mixing mechanism destabilises air over warm water, and
the increased turbulent mixing of momentum accelerates near-surface winds.
Conversely, cold SST suppresses the momentum mixing, decouples the
near-surface wind from wind aloft, and decreases the near-surface wind.
Another interesting feature seen in the wind field is the cold tongue of
Arctic Water in the southern Norwegian Sea (
Singularity exponents estimated from these mean satellite SST and wind speed fields, as shown in Fig. 4, provide a clearer and more consistent picture of the AF in the Norwegian Sea compared to the original scalar fields. The AF in the Norwegian Sea are characterised by negative singularity exponents. Previous studies using singularity exponent analysis also found similar results in frontal regions (e.g. Fig. 5 in Turiel et al., 2008). Regarding the SST, regions with negative singularity exponents correspond to regions with higher irregularity, i.e. to grid points with stronger gradient variations (see Figs. 3a and 4a for SST) than the neighbouring grid points (sharp transition). On the other hand, regions with positive singularity exponents are those where the gradient variations are weak. The frontal region at the Mohn Ridge (location 1 in Fig. 4a) and that associated with the cold tongue and the Faroe Current in the south-west Norwegian Sea (location 2) are the locations where the AF is most prominent. The imprint of East Icelandic Current can also be seen in Fig. 4a (location 3). The AF extending north-east from the cold tongue region is locked between the 1800 and 2400 m depth contour (location 4). The signature of the AF at location 4 is not very prominent compared to locations 1 and 2. Further north over the Vøring Plateau (location 5), the continuation of the AF is not captured in our analysis. Note that this is also a region of intense mesoscale eddy activity. The tracks of mesoscale eddies in the region obtained from Raj et al. (2016; updated time series up to 2016) is shown in Fig. S2. The figure shows dominance of mesoscale eddies across the AF at locations 4 and 5, more prominent at location 5. Eddies are known to contribute significantly to the total oceanic heat and salt transport by advective trapping (Dong et al., 2014), stirring and mixing (Morrow and Le Traon, 2012). Most importantly they can play an important role in the cross-frontal transport (Dufour et al., 2015). Eddy-induced mixing and cross-frontal transport can reduce the sharp distinction of the water masses at the AF in location 5 compared to the other four. This can be a reason for the absence of the AF signature in the singularity exponents map. Quantifying the impact of mesoscale eddies on the AF requires dedicated effort and is outside the scope of our study.
All the above SST-related features are also portrayed in the map of singularity exponents of the wind speed (Fig. 4b), estimated from the mean satellite wind speed map (shown in Fig. 3b). Regions with the sharpest changes in wind speed are consistent with those found in the SST map (Fig. 4a). In other words, regions where SST is found to experience sharp changes correspond well to those with large changes in wind speed, thereby confirming the role of strong momentum mixing along the frontal regions. The analysis is repeated using other remote sensing data sets (WindSAT SST, AVHRR SST, WindSAT wind speed and AMSRE wind speed) in order to examine whether the ocean fronts in the Norwegian Sea can be retrieved irrespective of the satellite data and sensor used. Singularity analysis on these satellite SST and wind speed data sets also illustrates similar AF signatures to those seen in Fig. 4 (not shown). Hence it can be stated that the satellite-derived wind speed and SST data reveal the frontal structure of the AF, irrespective of the sensor used. Thus, it is shown that singularity analysis is able to portray a much clearer representation of the AF compared to the original scalar fields.
Singularity exponents estimated from the climatological mean
(2002–2009)
Next, we assess the performance of the TOPAZ4 reanalysis results in reproducing the horizontal structure of the AF in the Norwegian Sea, using a longer period (1991–2015) than that of the satellite data (2002–2009). The surface signature of the AF estimated from TOPAZ4 SST output (Fig. 5a) agrees with that of the satellite SST data (Fig. 4a). As mentioned in Sect. 2.1, TOPAZ4 reanalysis assimilates measurements including along-track altimetry data, sea surface temperatures, sea ice concentrations and sea ice drift from satellites along with in situ temperature and salinity profiles. Hence, the similarities in Figs. 4a and 5a are not surprising, although not warranted due to the residual errors of assimilation. Nevertheless, the figures confirm that TOPAZ4 reanalysis data are able to reproduce the AF signature and indicates that the TOPAZ4 results can be used to study the subsurface part of the AF. Further analysis show that the horizontal structure of the AF in the deeper ocean is similar to the surface; most distinct along the Mohn Ridge and near the Faroe Current in the south-west Norwegian Sea (as shown by the TOPAZ4 potential temperature singularity exponent maps in Fig. 5b and c).
Singularity exponents estimated from TOPAZ
Singularity exponents estimated from TOPAZ
Next, in order to examine the seasonal variability of the AF in the Norwegian Sea, singularity analysis is applied to the seasonal climatologies of SST and potential temperature at 200 m (Fig. 6). The surface signature of AF is found to be stronger during winter (DJF; Fig. 6a), while the frontal signature is less evident during summer (JJA; Fig. 6b). These results agree with the analysis done using seasonal satellite SST data (not shown). Note that the analysis of the subsurface temperature fields (Fig. 6c and d) does not reproduce the seasonal variability of the surface frontal structure (Fig. 6a and b). Thus, it can be concluded that the seasonal variability of the frontal structure is limited to the surface and not found in the subsurface. This is likely to be associated with the surface heating of the ocean during summer that in turn reduces the surface horizontal temperature gradient, evidence of which is found as a thin layer of (upper 25 m) low-temperature gradient near the surface during summer (Fig. 7c), which is not present in winter (Fig. 7b), that disconnects the subsurface signatures of the front from the surface.
Temperature
The TOPAZ4 reanalysis, which successfully replicates the lateral variability
of the AF (Figs. 5–6), is further used to investigate the vertical structure
of the AF and its variability. The vertical structure of the AF is analysed
using a simple horizontal gradient method. The main reason for using a simple
horizontal gradient method to estimate the strength of the front is to
compare our results with previous studies (e.g. Piechura and Walczowski,
1995; Lobb et al., 2003). Furthermore, for this analysis the focus area is
limited to a single section (shown in Fig. 4a) taken across the Mohn Ridge,
where a strong frontal signature is found in the singularity exponents maps
(Figs. 4–6). Analysing the variability of the AF over the Mohn Ridge is also
important due to its proximity to Jan Mayen Island, which is well-known to be
an important breeding region inhabited by large colonies of seabirds.
Although higher up in the food chain, the variability of the AF may have an
influence on the birds through the impact on biology and fisheries of the
region. A similar section was used by Piechura and Walczowski (1995), to
study the AF across the Mohn Ridge using CTD data during the time period
1987–1993. Figure 7a, b and c show respectively the mean, winter and summer
climatology of potential temperature along the vertical cross-section at the
Mohn Ridge (location shown in Fig. 4a). To the east (left side of the
figure), the figure shows the warm Atlantic Water residing in the Lofoten
Basin, while the cold waters reside in the Greenland Sea
(right side of the figure). Note that these results agree with those
reported by Piechura and Walczowski (1995). The temperature gradient along
the vertical section delimits the AF at the Mohn Ridge (Fig. 7d). The
location of the core of the front along the section across the Mohn Ridge is
marked as red in Fig. 4a. At the location of the AF, a temperature change of
roughly 2
The impact of the large-scale atmospheric forcing in the Norwegian Sea
(Fig. 2) on the variability of the AF over the Mohn Ridge is investigated
next using composite analysis on monthly TOPAZ4 data. For this, composite
maps (Sect. 3.2) of the temperature across the Mohn Ridge during the positive
and negative phases of the NAO, EAP and SCAN are produced and the gradient in
mean temperature is estimated. The analysis shows that the location of the
core of the AF along the Mohn Ridge is not influenced by the large-scale
atmospheric variability (Figs. 8, S3). Here, the AF is defined as the region
with the maximum in temperature gradient, a classical definition used to
distinguish between two distinct water masses, in this case the Atlantic
Water and the Arctic Waters. The non-variability in the location of the
position of the AF further indicates that the location of the two water
masses does not alter. However, Fig. 8b shows a weakening of the AF during
the negative phase of the NAO (NAO
Gradient in the mean potential temperature (
Composite map of anomalous winds (vectors), during
Next, we investigate the link between the atmospheric circulation during
NAO
Composite of currents speed (cm s
While the majority of previous studies reported the positive relation of the
NAO with the speed of the NwASC (e.g. Skagseth and Orvik, 2002), our results
show that the NwAFC behaves in an opposite way. One of the possible
mechanisms by which a weakening in the cyclonic atmospheric circulation
during NAO
Composite map of anomalous BSFD, during
Graphical representation of the different processes associated
with the weakening of the AF during NAO
The discussion above explains the link between the variability in the
atmospheric circulation during NAO
Satellite-derived SST and wind speed data are good upper-level proxies for
the hydrographic signature of semi-permanent ocean fronts across the
Norwegian Sea. They reveal the frontal structure of the AF, irrespective of
the sensor used. The AF is found to be strongest over the Mohn Ridge and near
the Iceland Faroe Plateau. The study shows evidence of active air–sea
interaction along the AF. Compared to the original scalar fields (SST and
wind speed), a more precise picture of the AF in the Norwegian Sea is
obtained from their corresponding singularity exponents. The AF at the Mohn
Ridge extends from the surface down to 600 m depth. The seasonal variability
of the AF is limited to the surface and associated with the surface heating
of the ocean during summer that reduces the horizontal surface temperature
gradient. There is no seasonal variability in the location of the AF core
over the Mohn Ridge. For the first time, the role of the three dominant
atmospheric modes in the Norwegian Sea on the variability of the AF is
investigated. Out of the three, only the NAO influences the AF along the Mohn
Ridge. The study reveals a profound weakening of the AF over the Mohn Ridge
during NAO
All data sets used in this study are freely available.
TOPAZ model data are available via the CMEMS portal
(
In 1941, Kolmogorov (1941) introduced the concept of scale invariant
dissipations. Turbulence is characterised by an infinite amount of degrees of
freedom and thus it is impossible to fully characterise what happens at each
element of the fluid. However, the large number of degrees of freedom allowed
for a statistical approach, and thus turbulent flows could be characterised
by effective quantities. What Kolmogorov proposed was that there is a simple
relation between the energy dissipation at a scale
In 1985, Parisi and Frisch (1985) first introduced the concept of
multifractals to explain that in fact the properties of cascade variables
could be understood if we supposed that there is a hierarchy of fractal
components
The supplement related to this article is available online at:
RPR initiated the collaboration and designed the outline of the paper. SC and RPR were instrumental in the data analysis. RPR, SC, LB, AT and MP contributed to the interpretation of results. RPR led the writing of the paper with significant contributions from all co-authors, at all stages of the paper.
The authors declare that they have no conflict of interest.
This article is part of the special issue “The Copernicus Marine Environment Monitoring Service (CMEMS): scientific advances”. It does not belong to a conference.
This research is funded by the Copernicus Arctic MFC services. We also acknowledge the CPU provided by the Norwegian supercomputing project Sigma2 (nn2993k) and the storage space (ns2993k). Author Sourav thanks the Nansen Scientific Society for the support. We also thank Richard. E. Danielson for his helpful suggestions.
This research has been supported by the Copernicus Marine Services (grant no. 69).
This paper was edited by Ananda Pascual and reviewed by Arthur Capet and two anonymous referees.