Abstract

In the Nordic seas the Lofoten Basin is a region of high mesoscale activity. The generation mechanism and the conditions for the stability of a quasi-permanent vortex in the center of the Lofoten Basin are studied with a high-resolution ocean circulation model and altimeter data. The vortex and its generation mechanism manifest themselves by a pronounced sea surface height (SSH) signature and variability, which are found to be in agreement with altimeter data. The vortex results primarily from anticyclonic eddies shed from the eastern branch of the Norwegian Atlantic Current, which propagate southwestward. The large-scale bottom depression of the Lofoten Basin plays a crucial role for attracting anticyclones into the trough and for enabling the dynamical stability of the vortex. The water mass characteristics of the anticyclone lead to enhanced atmospheric interaction (heat loss) during wintertime. The cold water trapped in the upper part of the vortex preconditions convection in the following winter. This positive feedback mechanism tends to deepen convection progressively within the upper part of the vortex.

1. Introduction

The Lofoten Basin (LB) is the major heat reservoir in the Nordic seas and is therefore a region of large atmosphere–ocean interactions. It is separated from the Greenland Sea on its western side by the Mohns Ridge along which the polar front is located where both thermal and haline isolines rise steeply (Blindheim and Rey 2004). The largest mesoscale activity in the Nordic seas is also found in the Norwegian Sea, and especially in the LB (Poulain et al. 1996). Cross-frontal eddy transports of heat and salt are therefore the largest in this region (Rodionov 1992). At depth, the LB is characterized by anomalous large temperature and salinity values focused near the center of the basin, which can be found in most ocean atlases, for example, Boyer et al. (2005). Because climatologies resolve only large-scale distributions, the exact structure of this feature cannot be identified from climatological data alone. Russian field surveys in the 1970s and 1980s, specifically designed to resolve the synoptic scale of this anomaly, revealed a quasi-permanent anticyclonic eddy (Ivanov and Korablev 1995a, hereinafter IKa). The vortex is a doubly convex lens of a usually less-than-50-km horizontal scale, which is enhanced during wintertime and relaxed during summer. IKa concluded that the regeneration of the vortex is due to the penetration of surface water during convection periods. The particular location of the vortex and mechanisms for the drift are described by Ivanov and Korablev (1995b, IKb hereinafter) as a topographically controlled cyclonic circulation along f /H contours.

Here, we use a high-resolution general circulation model of the Nordic seas (Köhl et al. 2007a) to investigate mechanisms for the generation and stability of the vortex. This analysis is based on a simulation of the circulation in the Nordic seas during the last 13 yr. Theoretical considerations are explored to investigate mechanisms for the generation and conditions for the stability of the anticyclone. Although convection contributes to the regeneration of the anticyclone, other mechanisms are found to be more relevant. Likewise, the importance of topographic control for the selection of the location of the vortex can be confirmed, but a more detailed investigation revealed that the mechanism differ from that suggested by IKb. The aim of the paper is to revisit dynamical aspects of the anticyclone in the center of the LB.

The paper is organized as follows. Section 2 provides a description of the numerical model and section 3 describes the characteristics of the vortex from the simulation. The relevance of convection is described in section 4. Generation mechanisms and conditions for stability are presented in section 5 and 6, respectively. In section 7 we describe the selection of the position of the vortex. Section 8 provides a summary and concluding remarks.

2. The model

The numerical simulation is based on the Massachusetts Institute of Technology (MIT) ocean general circulation model (Marshall et al. 1997a, b), configured for the region of the eastern subpolar North Atlantic Ocean and Nordic seas from 51° to 78°N. It is nested into the global 1° state estimation of the Estimation of the Circulation and Climate of the Ocean (ECCO) consortium (Köhl et al. 2007b). The grid is horizontally isotropic with a zonal resolution of 1/10° and has 30 levels in the vertical. The model was started from rest and climatological temperature and salinity (Levitus and Boyer 1994; Levitus et al. 1994). Over the period from 1992 to 2004 the simulation was forced by the daily atmospheric state obtained from the National Centers for Environmental Prediction–National Center for Atmosphere Research (NCEP–NCAR) reanalysis project using the bulk formula according to Large and Pond (1981, 1982. The model formulation includes a parameterization for vertical mixing by the K-profile parameterization (KPP) scheme of Large et al. (1994) and a dynamic/thermodynamic sea ice model of Zhang and Rothrock (2000). Additional details of this simulation can be found in Köhl et al. (2007a). Using the same numerical code, a process model was set up covering only the region from 66.8° to 71.6°N, and from 4.45°W to 11.5°E. It was run with various idealistic initial conditions described below. No surface forcing or open boundaries were applied.

3. Quasi-permanent vortex

The North Atlantic Current (NAC) enters the Nordic seas in three major current branches (see Fig. 1). These are the Irminger Current, which enters west of Iceland, and the Norwegian Atlantic Current (NwAC), which is separated into two branches and which enters east of Iceland. The eastern part follows the coastal shelf of Norway closely. The western part of the NwAC is also topographically guided and represents the jet in the polar front. It initially follows the slope of the Vøring Plateau toward Jan Mayen Island, and continues afterward along the Mohns Ridge toward the Fram Strait (Orvik and Niiler 2002). The Norwegian Coastal Current (NCC) is located even farther east on the Norwegian shelf.

Fig. 1.

Schematic of the major pathways of the near-surface water in the Nordic seas. The major currents are the NwAC, which carries warm and salty water from the North Atlantic into the Nordic seas, and the East Greenland Current (EG), which carries polar water that entered through Fram Strait (FS). The NCC carries relatively freshwater of Baltic origin on the Norwegian shelf northward. The four major basins are the LB, Norwegian Basin (NB), Greenland Basin (GB), and Iceland Basin (IB).

Fig. 1.

Schematic of the major pathways of the near-surface water in the Nordic seas. The major currents are the NwAC, which carries warm and salty water from the North Atlantic into the Nordic seas, and the East Greenland Current (EG), which carries polar water that entered through Fram Strait (FS). The NCC carries relatively freshwater of Baltic origin on the Norwegian shelf northward. The four major basins are the LB, Norwegian Basin (NB), Greenland Basin (GB), and Iceland Basin (IB).

The water masses in the LB are characterized by warm and saline Atlantic water that occupies the upper 600–700 m and reaches eastward to the polar front. In this layer salinity values are higher than 35 psu and temperatures exceed 3°–4°C. Orvik (2004) describes the deepening of the Atlantic water in the Lofoten Basin and presents a mechanism based on the presence of the deep countercurrent over the Mohns Ridge. Below this layer, in the temperature range from −0.5° to 0.5°C, low-salinity water from the Greenland Basin intrudes the LB and the water farther below, which is characterized by a weak salinity maximum, is of Arctic origin, according to Blindheim and Rey (2004). Very low salinity values are found in the upper 200 m near the coast of Norway. This second frontal structure is associated with the eastern NwAC and the NCC, which carries very fresh water originating from the Baltic Sea and Norwegian Fjords northward (Björk et al. 2001). The NCC is described by Pedersen et al. (2005) as a baroclinically unstable current that forms cyclonic and anticyclonic eddies. The LB is thus located between two frontal structures that generate variability.

In Fig. 2 we compare the rms sea surface height (SSH) variability calculated from our 1992–2004 simulation with the variability based on maps from altimeter data provided by Archiving, Validation and Interpretation of Satellite Oceanographic data (AVISO; online at http://www.aviso.oceanobs.com) on a ⅓° grid. Both calculations reveal high variability in the LB with a clear maximum at the center of the basin, and maximum rms values reaching from 9 cm for the model to 11 cm based on altimeter data. Some of the smaller-scale structures are missing in the altimeter-based variance, most likely because of the usage of too large a decorrelation length scale during the optimal interpolation. The modeled maximum variability is about 20% lower than in the observations (possibly because of either errors in the atmospheric forcing or the lack of resolution), but the center of the maximum matches the observed position. The observed seasonal component of the variability has an amplitude of about 8 cm (Jakobsen et al. 2003), which explains about 50% of the observed rms SSH variability. The authors attribute, in agreement with the analysis of Isachsen et al. (2003), the seasonal part of the observed signal to the seasonal spinup of the gyres resulting from changing wind stress. In the model, the amplitude of the seasonal part amounts to only 4 cm, which is partly due to the underestimation of modeled variability, but more importantly is due to a partially missing steric response in the model that is, because of the Boussinesq approximation, volume rather than mass conserving. After removal of the basin mean from the altimeter data (the part that is not simulated by the model and that has no dynamical consequences) the amplitude of the seasonal cycle is reduced to less than 6 cm, which explains about 5 cm and thus only about 50% of the observed rms variability. Although the wind-induced gyre response certainly also acts on time scales other than the seasonal, the study by Isachsen et al. (2003) shows that the seasonal variability is not entirely explained by wind stress variations. A large part of the variability observed in the LB thus remains unexplained. A similar ratio holds also for the simulation.

Fig. 2.

Rms SSH variability (top) of the model and (bottom) from altimeter data maps (cm). The mean over the model domain (51°–78°N, 47°W–17°E) was subtracted before the calculation in order to remove the seasonal variability in the data resulting from steric expansion. The bathymetry is overlaid with a contour-level interval (CI) of 500 m.

Fig. 2.

Rms SSH variability (top) of the model and (bottom) from altimeter data maps (cm). The mean over the model domain (51°–78°N, 47°W–17°E) was subtracted before the calculation in order to remove the seasonal variability in the data resulting from steric expansion. The bathymetry is overlaid with a contour-level interval (CI) of 500 m.

The circulation of the model reveals, as in IKa, a very strong anticyclonic eddy in the center of the LB, which is often surrounded by other, yet only transient, eddies. A section along 69.8°N of potential density shows the structure both as a 1998–2003 mean (Fig. 3a) and snapshot (Fig. 3b). The vortex has the structure of a doubly convex lens with the center at about 600-m depth and a diameter of about 30 km. The vortex is mainly a thermal structure (heton) with a smaller haline contribution. In terms of density changes the ratio is about two. Because of the lenslike structure of the hydrography, the vortex shows a subsurface maximum of the associated velocity field also at about 600 m. The characteristics of the simulated vortex are in good agreement with the hydrographic measurements of IKa. We will show in the following that higher temperature and salinity below 500 m are a consequence of the anticyclonic origin of the eddy. Lower temperature and salinity values above 500 m are due to wintertime convection. The SSH signature (Fig. 3f) of the vortex is on the order of 20 cm, with almost equally strong ambient cyclones. Because the vortex is interacting with the surrounding vortices and is also advected with the mean cyclonic circulation in the LB (IKb), most of the SSH variability in this basin can be attributed to the movement of this anticyclone.

Fig. 3.

Potential density along 69.8°N averaged (a) over 1998–2003 and (b) in August 2003. Hydrographic sections along 69.8°N for (c) temperature (°C), (d) salinity (psu), and (e) meridional velocity (m s−1) for August 2003, as well as (f) SSH (cm) for the same month.

Fig. 3.

Potential density along 69.8°N averaged (a) over 1998–2003 and (b) in August 2003. Hydrographic sections along 69.8°N for (c) temperature (°C), (d) salinity (psu), and (e) meridional velocity (m s−1) for August 2003, as well as (f) SSH (cm) for the same month.

4. The relevance of convection

The Lofoten Basin is collocated with an area of enhanced wintertime convection that mixes water down to depths of around 400 m (Kara et al. (2002; see also Fig. 4a). In the LB, wintertime mixed layers (MLs) even reach down to 600 m (Nilsen and Falck 2006). Over the course of the integration it becomes obvious that deeper mixed layer depths (MLDs) are observed in the interior of anticyclones (Fig. 4b) and that in particular the LB vortex shows progressively deeper mixed layers (Fig. 4d). Although the deepening in the model is mainly considered as part of the spinup process of the model, it gives additional insight into the dynamics of the vortex. IKa describes a seasonal dependence of the size of the anticyclonic vortex in which the vortex shrinks in winter to a Rossby radius of about 10 km and expands during late summer to 5–7 times the Rossby radius, although this seems to be not particularly supported by their data. They argue that eddy fluxes will cause a gradual relaxation of the structure and propose a regeneration process based on convection.

Fig. 4.

MLD (m) calculated as the depth where the density difference σ0(z) − σ0(z = 0) reaches ⅛ sigma units. MLD of snapshots for the years (a) 1992, (b) 1994, and (c) 2002. Contour lines of the bathymetry are overlaid using a CI of 400 m. (d) Maximum MLD in the LB (69°–70.5°N, 0°–7°E) as time series.

Fig. 4.

MLD (m) calculated as the depth where the density difference σ0(z) − σ0(z = 0) reaches ⅛ sigma units. MLD of snapshots for the years (a) 1992, (b) 1994, and (c) 2002. Contour lines of the bathymetry are overlaid using a CI of 400 m. (d) Maximum MLD in the LB (69°–70.5°N, 0°–7°E) as time series.

As described by Send and Marshall (1995), in a convection chimney the cooled water mixes with the water below and the subsequent geostrophic adjustment implies a net downward motion. The isopycnals below the mixing chimney move downward and establish an anticyclonic movement at depth. The colder water in the upper part implies cyclonic movement. After surface restratification during summer, a lenslike hydrographic structure remains. These lenslike vortices were observed in the Greenland Sea and are known to live long (Gascard et al. 2002).

Although this mechanism is in principle able to explain a lenslike structure as observed in the LB and does certainly contribute to the regeneration as argued by IKa, we will present modeling evidence that this is not the most relevant mechanism for establishing the anticyclone. It appears from Fig. 3 that the existence of anticyclones is a prerequisite rather than a consequence for enhanced convection, as during the first winter after initialization; from the Levitus temperature and salinity a visible vortex is not yet formed and the convective region is of a much larger scale. The first time the anticyclone appears in the simulation at the center of the LB is at the end of September in the same year. This anticyclone does not yet show the lenslike hydrographic structure. It has moved into the center from farther east and has a strong positive SSH signature. Note that the fully developed vortex in Fig. 3 maintained the positive SSH. In contrast, a negative SSH is typical for a convectively formed lens.

Anticyclones in the LB are formed from water that is warmer but also saltier than the environment, and especially the environmental cyclones. The former characteristic leads to a larger heat loss during wintertime and thus a larger buoyancy flux that promotes convection in anticyclones. However, because the density in the upper part of an anticyclone is lower than in a cyclone, the static stability is larger in the anticyclone (a fact that suppresses convection). The balance between these two counteracting factors will determine whether convection is deeper or shallower in anticyclones. In the following, a very simple model is regarded in order to decide upon whether cyclones or anticyclones favor vertical mixing in the LB. According to Nilsen and Falck (2006) most of the month-to-month deepening of the MLD can be explained by atmospheric forcing whereby, especially in wintertime, the buoyancy forcing explains more than ¾ of the MLD variations. We will therefore disregard in our simplified consideration all advective and wind-driven effects.

The buoyancy flux is dominated by the heat flux, which is to the first order assumed to be proportional to the temperature difference between the ML and the atmosphere TmlTa. This is apart from its dependency on the atmospheric wind speed UA, which is exact for the sensible heat flux and approximately true for the latent heat flux, because it is strongly correlated in the LB (r > 0.9 for the NCEP data). Shortwave radiation is not regarded because its effect on buoyancy is (except for the nonlinearity of the density equation) the same for cyclones and anticyclones. The exclusion of longwave radiation, which additionally favors convection in anticyclones, biases our assessment of stronger convection in cyclones. We now assume a linear temperature profile with constant temperatures Tml and To in the mixed layer and below thermocline, respectively. The latter is assumed to be the same for both types of vortices. If the temperature profile in the thermocline of thickness ΔH is then assumed to be linear, ∂T/∂z = (TmlTo)/ΔH, the change in MLD with time t resulting from surface cooling reads

 
formula

If Ta is smaller than To then the change of the MLD will be larger for the smaller Tml, and therefore convection will be deeper for the cold, cyclonic eddy. For To = −1°C taken from the model profiles (see Fig. 9), this is true on average for only 10 days yr−1 during the period of 1992–2004, as opposed to 191 days at which convection in anticyclones is favored.

Fig. 9.

Temperature and salinity profiles in the LB from the 1992 to 2004 simulation for the background (dotted), LB vortex (thick), anticyclone (dashed), and cyclone (dash–dotted).

Fig. 9.

Temperature and salinity profiles in the LB from the 1992 to 2004 simulation for the background (dotted), LB vortex (thick), anticyclone (dashed), and cyclone (dash–dotted).

Because salinity is higher near the surface the presence of salinity reduces static stability and enhances the MLD changes. Higher salinity in the anticyclone therefore additionally promotes larger MLD. The effect can be included by regarding density gradients rather than temperature gradients. However, a similar consideration with the accordingly modified equation shows that the additional effect is small, increasing the number of days at which convection is favored in anticyclones only by 2. Altogether, this simple model is not meant to provide an exact estimate of the number of days that favor larger MLD in anticyclones. The main purpose is to show why MLDs appear to be deeper in anticyclones than in cyclones.

The general preference of wintertime deepening of the MLD in anticyclones is, however, not sufficient to explain either the exceptionally large MLD in the interior of the LB vortex or why the MLD progressively deepens over the course of the integration. The convectively mixed water is kept inside the eddy rather than being carried away with the mean current (the NAC). A well-mixed area is therefore present in the following winter and preconditions convection at the location of the eddy; consequently, the convection depths are progressively enhanced during the following winters. The preservation of the cold water in the upper part of the vortex accounts for the lenslike structure and the progressive deepening of the ML involves a downward shift of the center of the vortex.

Although we find deeper MLD in anticyclones, our simulation does not support a clear seasonal dependency of the radius R if the vortex is characterized by a Gaussian shape {exp[−(x2 + y2)/2R2]}. The estimation of the horizontal scale by IKa depends on the temperature difference between surface and core, which basically is chosen to be sensitive to horizontal gradients of SST only. Using the same criteria we find values R = 7–71 (8–48) km, with a standard deviation of 13 (8) km. The amplitude of the seasonal is 2 (4) km and the phase is 187 (127) days (values in parentheses are calculations based on SSH values). In summary, the seasonal cycle accounts for less than 30% of the variability with a maximum occurring during late spring. Although our values are not completely incompatible with the findings of IKa, the time lag between maximum convection and maximum vortex size suggests a different mechanism, which we will discuss in the following section.

5. The generation mechanism

Poulain et al. (1996) describes the LB as a region with maximum eddy activity but just average residence time of surface drifters. The Eulerian mean circulation calculated from drifter data indicate, in accord with f /H contours, a general cyclonic circulation pattern in the LB. Despite the existence of the permanent anticyclone in the center of the basin, they report only a few long-lived mesoscale features in the LB. During spinup from rest and for Levitus T and S, we observe from our simulation that the eastern NwAC and NCC become almost immediately unstable and produce strong anticyclonic (as well as cyclonic) eddies within less than a month. In accord with planetary β, the eddies tend to move southwestward. Figure 5 compares the propagation of SSH anomalies along 69.8°N from modeled SSH with those from altimeter maps. Both plots consistently show the largest variability between 2° and 6°E associated with the movement of the quasi-permanent anticyclone, and a general westward propagation of, especially, the anticyclones. An Eulerian correlation analysis performed on the drifter data by Poulain et al. (1996) also indicates a westward propagation (with a propagation speed of about 10–20 cm s−1) of the eddies within the Norwegian Sea and LB. The westward propagation speed of longer-living vortices estimated from the Hovmöller diagram is generally on the order of 2 cm s−1, but maximum propagation speeds of more than 10 cm s−1 are observable from tracing individual eddies.

Fig. 5.

Hovmöller diagram of SSH along 69.8°N from (top) simulation and (bottom) mapped altimeter data (cm) shown only for the years 1995–2000 for clarity. The spatial mean along 69.8°N was subtracted in order to remove the steric effect of seasonal heating from the altimeter data, and the time mean was removed from the model data because the altimeter data represent the anomalies to the time mean only. The westward propagation of 2 cm s−1 is indicated as a green line (faster propagation and less clear for the altimeter data). The sampling rate for the simulated SSH is 3 and 7 days for the altimeter data.

Fig. 5.

Hovmöller diagram of SSH along 69.8°N from (top) simulation and (bottom) mapped altimeter data (cm) shown only for the years 1995–2000 for clarity. The spatial mean along 69.8°N was subtracted in order to remove the steric effect of seasonal heating from the altimeter data, and the time mean was removed from the model data because the altimeter data represent the anomalies to the time mean only. The westward propagation of 2 cm s−1 is indicated as a green line (faster propagation and less clear for the altimeter data). The sampling rate for the simulated SSH is 3 and 7 days for the altimeter data.

Over the course of the first 6 months anticyclonic eddies are propagating southwestward from the east into the trough of the LB and establish a quasi-stationary anticyclone that is surrounded by cyclones and weaker anticyclones. The latter merge occasionally with the central anticyclone, creating a stronger vortex. One merging process is shown in Fig. 6 at a later stage of the integration after the vortex has spun up. On average, there are about three to four merging processes a year that are relatively uniformly occurring, with a slight bias toward the months from February through May and no occurrences in November and December. A larger number of mergers in this period is consistent with a NwAC that is, in many years, strongest in January–February, because a stronger current is more unstable and will shed more eddies toward the west. The trough of the LB thus serves as a catching area for anticyclonic eddies, a fact that promotes the merging of vortices. It is noteworthy to mention that a parallel run with a coarser ¼° twin model, which shows considerably less eddy activity, develops only a slight indication of an anticyclone in the center of the LB. The Atlantic water is accordingly significantly less deep in the LB, and for the remaining deepening the mechanism described by Orvik (2004) might be important.

Fig. 6.

Merging of anticyclones in the LB. Contours of temperature at 1000 m over the bottom topography from four consecutive snapshots each 3 days apart, starting with 4 May 2001. Only the contour lines of 0.5°, 1°, 2°, and 3°C are shown for clarity of the plot.

Fig. 6.

Merging of anticyclones in the LB. Contours of temperature at 1000 m over the bottom topography from four consecutive snapshots each 3 days apart, starting with 4 May 2001. Only the contour lines of 0.5°, 1°, 2°, and 3°C are shown for clarity of the plot.

Figure 7 shows one of the rare breakup processes. The separated vortices remerge during the following month. The stability considerations in the following section suggest that the size of the vortex is crucial for stability and that a large vortex might become unstable. It is therefore possible that the vortex grows because of a merger with ambient anticyclones and may reach a maximum size beyond which it becomes unstable. However, because only one true breakup (only small parts became detached at other times) was observed, this instability mechanism could not be verified.

Fig. 7.

Breakup of anticyclones in the LB. Contours of temperature at 1000 m over the bottom topography from four consecutive snapshots each 3 days apart, starting with 26 Nov 2003. Only the contour lines of 0.5°, 1°, 2°, and 3°C are shown for clarity of the plot.

Fig. 7.

Breakup of anticyclones in the LB. Contours of temperature at 1000 m over the bottom topography from four consecutive snapshots each 3 days apart, starting with 26 Nov 2003. Only the contour lines of 0.5°, 1°, 2°, and 3°C are shown for clarity of the plot.

6. Stability of the vortex

In this section we show that the stability of the vortex is caused by the specific topographic conditions in this basin. As the mean flow, vortices are subjected to barotropic and baroclinic instability. Observed eddies are often found to be more stable than simple instability considerations suggest. Attempts to resolve this discrepancy include the consideration of deep flow (e.g., Dewar and Killworth 1995) or bottom topography (Nycander and LaCasce 2004). Both effects can either stabilize eddies or at least weaken the instability. In the framework of a quasigeostrophic model, Benilov (2005) provided a more quantitative study in which he demonstrated that bottom elevation tends to stabilize cyclones, whereas bottom depressions stabilize anticyclones. Relatively small topographic changes are shown to provide considerable stabilization effects.

The large spatial scale of the depression, which the LB represents, thus has the tendency to stabilize anticyclonic eddies and destabilize cyclonic eddies. For the stability analysis, Benilov (2005) regards Gaussian vortices over a Gaussian-shaped topography in a two-layer ocean. After nondimensionalization with the Rossby radius Ld, the stability depends on the nondimensional radius of the eddy re, the bottom slope (also expressed by a radius rt), the ratio of the upper layer to the lower layer thickness ɛ = Hupper/Hlower, and the topographical parameter δ = foLdΔlower/Uupper/Hupper. Here fo is the Coriolis parameter, Δlower is a characteristic topographic variation, and Uupper is a typical upper layer velocity. From Fig. 3 we estimate Hupper = 800 m and Hlower = 2400 m, and thus ɛ = ⅓. The stability curve of second azimuthal mode, which in most cases turned out to be the most unstable, is shown in Fig. 8, which corresponds to Fig. 1 in Benilov (2005).

Fig. 8.

Marginal stability curves from Benilov (2005) for a Gaussian vortex over a Gaussian topography, and the second azimuthal mode, which is the most unstable mode in most cases. Here ɛ is the depth ratio of the ocean and re is the radius of the vortex. Stability curves are shown for the topography parameter δ = −1, 0, 1. (a) rt = 2 and (b) rt = 1. The range of possible anticyclones in the LB is marked in (a).

Fig. 8.

Marginal stability curves from Benilov (2005) for a Gaussian vortex over a Gaussian topography, and the second azimuthal mode, which is the most unstable mode in most cases. Here ɛ is the depth ratio of the ocean and re is the radius of the vortex. Stability curves are shown for the topography parameter δ = −1, 0, 1. (a) rt = 2 and (b) rt = 1. The range of possible anticyclones in the LB is marked in (a).

The first baroclinic Rossby radius Ld was calculated according to Emery et al. (1984) from vertical profiles of the buoyancy frequency N2 = gzρ/ρo, with g being the gravitation constant and ρ being the potential density. From time mean profiles of the model averaged over the region of 69°–7°N, 0°–8°W, a Rossby radius of Ld = 27.8 km was estimated. For the calculation of the radius of the vortex, the temperature was scaled with the maximum and average temperature over the region of 69°–70.5°N, 1°–6°W. Because the scaled temperature curve is very well approximated by a Gaussian function, the radius could be calculated from the area of values larger than exp(−0.5). The radius shows random fluctuations [σre(200 m)=11 km and σre(1000 m)=5 km] and a slight decrease in size over time from approximately 35 to 30 km at 200 m and from 30 to 28 km at 1000 m. After nondimensionalization with the Rossby radius, the vortex radius lies within a range of re = 1.08–1.26, which is, according to Figs. 8a,b, at ɛ = ⅓ very close to the neutral stability curve without topographic effect. For larger values of re the vortex is in the range of baroclinic instability.

The topographic radius was fitted by overlaying the Gaussian shape and the real topography by eye to rt = 2.2° = 85.2 km, which yields a nondimensionalized value of 3.1. A typical upper layer velocity of Uupper = 0.2 m s−1 and of about Δlower = 500 m yields δ = 1.3. Comparing Figs. 1a, b of Benilov (2005) suggests that a larger rt as well as larger δ value will extend the range of stability probably beyond re = 1.5, which should mean that the presence of the large-scale topographic depression in the LB should stabilize most anticyclones in this region. Moreover, the stability curve shows that anticyclones are only marginally stable or weakly unstable elsewhere.

To verify the results obtained in the framework of the simplified model of Benilov (2005), several experiments were performed with the simpler concept model covering only the domain of the LB. The model was initialized with a Gaussian vortex of typical size and hydrography of the anticyclones in the LB. The far field was taken from background profiles in the LB. Profiles of the hydrographic conditions of cyclones, anticyclones, and the far field are taken from the respective positions and are shown together with a profile of the quasi-permanent vortex in Fig. 9.

Figure 10a shows the evolution of the anticyclone over 200 days if no bottom topography is imposed. In accord with the theoretical consideration, the initial vortex breaks up into several vortices after 200 days, but the presence of a Gaussian bottom depression as estimated above from the realistic topography stabilizes the vortex (Fig. 10b). Likewise, the real bottom topography is also able to stabilize the vortex (not shown), and is thus a necessary condition for the presence of the anticyclone in the LB.

Fig. 10.

Stability investigation with the concept model. The temperature anomaly is shown with respect to the area mean in 200 m of the first time step (red) and after 200 days (blue), together with the trajectory of the movement of the center (green). (a) Flat bottom; the anticyclone is initialized with temperature and salinity profiles of Fig. 9 and geostrophic velocities. The horizontal distribution is Gaussian. (b) Initialized the same as in (a), but with Gaussian bottom topography (CI = 200 m). (c) The merging of two anticyclones is initialized as in (a). (d) Anticyclone is the same as in (b), but with four ambient cyclones initialized with the corresponding profiles of Fig. 9.

Fig. 10.

Stability investigation with the concept model. The temperature anomaly is shown with respect to the area mean in 200 m of the first time step (red) and after 200 days (blue), together with the trajectory of the movement of the center (green). (a) Flat bottom; the anticyclone is initialized with temperature and salinity profiles of Fig. 9 and geostrophic velocities. The horizontal distribution is Gaussian. (b) Initialized the same as in (a), but with Gaussian bottom topography (CI = 200 m). (c) The merging of two anticyclones is initialized as in (a). (d) Anticyclone is the same as in (b), but with four ambient cyclones initialized with the corresponding profiles of Fig. 9.

7. Position of the vortex

Although the stability analysis of Benilov (2005) is able to explain why the vortex remains stable, it is neither able to explain why it stays in the center of the LB nor why other anticyclones are attracted into the basin and merge with the vortex. The planetary β effect causes anticyclones to drift southwestward and cyclones to propagate to the northwest. Likewise, the topographic β also affects the movement of vortices. According to Carnevale et al. (1991b), cyclones will climb up the slope in an anticyclonic spiral relative to the center of a seamount and anticylones will descent toward the center of a bottom depression in a cyclonic spiral. They verified their prediction by rotation tank experiments. With this simple mechanism anticyclones will be caught in a trough, for example, at the center of a basin. The prerequisite for topographic β to be felt by the vortex is that the horizontal scale of the vortex be small in comparison with the scale of the depression. In other words, the same condition as that of a large-scale depression, which was necessary for the stability, is required for the LB to become attractive for anticyclones. As in the rotation tank experiments the existence of a spiral path is difficult to verify, and the path of the vortex shown in Fig. 10b is only consistent with a cyclonic spiral but does not provide independent evidence.

From theoretical considerations, for example, Melander et al. (1988) have predicted that two vortices of diameter 2R will merge when their separating distance is smaller than 3.2R. For larger distances two like-signed vortices will circle around a common center without getting closer. The critical distance was verified in laboratory experiments for a pair of anticyclones by Griffith and Hopfinger (1987) who found that cyclones may also merge for much larger initial distances. Because all anticyclones are attracted by the center of the depression, their separation to each other will naturally be reduced below the critical distance and they will merge. Carnevale et al. (1991a) in fact showed that the parabolic depression of the surface (equivalent to a seamount resulting from the reduction of the water depth) in rotating tank experiments will cause the cyclones to move to the center and explain the anomalous vortex merger for larger than the critical distance. The process model was initialized with two anticyclones separated by a distance of 6 times their radius. Sizes and water mass characteristics (see Fig. 9) are typical for the LB. Figure 10c shows that two anticyclones will merge in the presence of a bottom depression similar to that found in the LB for larger than the critical distance.

According to Carnevale et al. (1991b) and Fig. 10b, the anticyclone should spiral on a cyclonic path toward the center of the basin. Because of disturbances by ambient eddies in the real ocean the path will be more random, but a cyclonic movement should be predominant. In Fig. 11 the averaged displacement vector field is shown, which was calculated from the trajectory of the movement of the vortex center. The clear anticyclonic movement contradicts the above explanation, but also disagrees with the findings of IKb who explain the movement of the center by the advection with the mean current that follows f /H contours and should thus be cyclonic. Their finding is based on very few observations of the vortex position, and random interactions with surrounding eddies may cause a cyclonic path during the time of observation. Near the center of the basin the drift resulting from topographic β is very small and the cyclonic gyre in the LB is weak; therefore, other effects may dominate. From the mean SSH (see also Fig. 3f) it appears that the anticyclone is usually surrounded by cyclones. The interaction with these cyclones will guide the central anticyclone to follow, on average, an anticyclonic path. In Fig. 10d the effect of surrounding cyclones is accordingly shown to lead to an anticyclonic movement of a vortex in the center. The cyclones are probably stabilized by the topographical rises that define the margin of the LB. Moreover, the same mechanism that attracts the anticyclone into the center of the LB will also keep the cyclones in the periphery of the LB.

Fig. 11.

Averaged movement of the center (° month−1) over the relative frequency distribution of the center locations. The velocity plot was calculated from the displacement vectors of the eddy center over the last 6 yr of the simulation. Only vectors for relative frequency values larger than 0.1 were plotted.

Fig. 11.

Averaged movement of the center (° month−1) over the relative frequency distribution of the center locations. The velocity plot was calculated from the displacement vectors of the eddy center over the last 6 yr of the simulation. Only vectors for relative frequency values larger than 0.1 were plotted.

8. Conclusions

We revisited the properties and dynamics of a quasi-permanent vortex in the LB that already was formerly described by IKa and IKb in considerable detail. Our eddy-resolving simulation of the Nordic seas suggests a different formation and regeneration mechanism than that described by IKa. In contrast to the suggested regeneration through surface convection, the vortex is found to originate primarily from anticyclones that were originally created by the instability of the NwAC and that subsequently propagate into the LB. Convection is found to play mainly an important role in setting the conditions of the upper part of the lenslike hydrographic structure. Regeneration of the vortex is in agreement with rotation tank experiments through the anomalous merger of anticyclones facilitated by the large-scale bottom depression of the LB. This large-scale topographic condition attracts anticyclones in the center of the LB and therefore also explains the selection of the position of the vortex near the center of the basin. The application of results from a recent theoretical stability investigation by Benilov (2005) suggests that the vortex, because of its size and hydrographic conditions, is only marginally stable, but that the whole range of existing vortex diameters is stabilized by the large-scale bottom depression of the LB. Moreover, the stability range suggests that the vortex may grow through merger to a size where it is no longer stable and breaks up. One such breakup was simulated, although the reason could not be investigated. Other than that described by IKb, the vortex center is not found to follow a cyclonic path guided by the mean circulation, which is described by f /H contours. However, the calculated mean anticyclonic movement also contradicts the suggested cyclonic spiral induced to topographic β. It is suggested that these mechanisms are not strong enough and that the interaction with ambient cyclones determines the path to be anticyclonic on average. Altogether, the LB is a region where many theoretical concepts apply to facilitate the existence and stability of the vortex and where these concepts can be tested in a large-scale environment.

Acknowledgments

Detlef Quadfasel and an anonymous reviewer provided valuable comments. This work was supported by the Deutsche Forschungsgemeinschaft in the Sonderforschungsbereich 512 project. The integration of the model simulations was performed at the German high-performance computing center DKRZ. SSH maps were produced by Ssalto/Duacs as part of the Environment and Climate EU Enact project (EVK2-CT2001-00117) and distributed by Aviso, with support from CNES.

REFERENCES

REFERENCES
Benilov
,
E. S.
,
2005
:
Stability of a two-layer quasigeostrophic vortex over axisymmetric localized topography.
J. Phys. Oceanogr.
,
35
,
123
130
.
Björk
,
G.
,
B. G.
Gustafsson
, and
A.
Stigebrandt
,
2001
:
Upper layer circulation of the Nordic seas as inferred from the spatial distribution of heat and freshwater content.
Polar Res.
,
20
,
161
168
.
Blindheim
,
J.
, and
F.
Rey
,
2004
:
Water-mass formation and distribution in the Nordic seas during the 1990s.
ICES J. Mar. Sci.
,
61
,
846
863
.
Boyer
,
T.
,
S.
Levitus
,
H.
Garcia
,
R. A.
Locarnini
,
C.
Stephens
, and
J.
Antonov
,
2005
:
Objective analyses of annual, seasonal, and monthly temperature and salinity for the world ocean on a ¼° grid.
Int. J. Climatol.
,
25
,
931
945
.
Carnevale
,
G. F.
,
P.
Cavazza
,
P.
Orlandi
, and
R.
Purini
,
1991a
:
An explanation for anomalous vortex merger in rotating-tank experiments.
Phys. Fluids
,
3A
,
1411
1415
.
Carnevale
,
G. F.
,
R. C.
Kloosterziel
, and
G. J. F.
van Heijst
,
1991b
:
Propagation of barotropic vortices over topography in a rotating tank.
J. Fluid Mech.
,
223
,
119
139
.
Dewar
,
W. K.
, and
P. D.
Killworth
,
1995
:
On the stability of oceanic rings.
J. Phys. Oceanogr.
,
25
,
1467
1487
.
Emery
,
W. J.
,
W. G.
Lee
, and
L.
Magaard
,
1984
:
Geographic and seasonal distributions of Brunt–Väisälä frequency and Rossby radii in the North Pacific and North Atlantic.
J. Phys. Oceanogr.
,
14
,
294
317
.
Gascard
,
J-C.
,
A. J.
Watson
,
M-J.
Messias
,
K. A.
Olsson
,
T.
Johannessen
, and
K.
Simonsen
,
2002
:
Long-lived vortices as a mode of deep ventilation in the Greenland Sea.
Nature
,
416
,
525
527
.
Griffith
,
R. W.
, and
E. J.
Hopfinger
,
1987
:
Coalescing of geostrophic vortices.
J. Fluid Mech.
,
178
,
73
97
.
Isachsen
,
P. E.
,
J. H.
LaCasce
,
C.
Mauritzen
, and
S.
Häkkinen
,
2003
:
Wind-driven variability of the large-scale recirculating flow in the Nordic seas and Arctic Ocean.
J. Phys. Oceanogr.
,
33
,
2534
2550
.
Ivanov
,
V.
, and
A.
Korablev
,
1995a
:
Formation and regeneration of the pycnocline lens in the Norwegian Sea.
Russ. Meteor. Hydrol.
,
9
,
62
69
.
Ivanov
,
V.
, and
A.
Korablev
,
1995b
:
Interpycnocline lens dynamics in the Norwegian Sea.
Russ. Meteor. Hydrol.
,
10
,
32
37
.
Jakobsen
,
P. K.
,
M. H.
Ribergaard
,
D.
Quadfasel
,
T.
Schmith
, and
C. W.
Hughes
,
2003
:
Near-surface circulation in the northern North Atlantic as inferred from Lagrangian drifters: Variability from the mesoscale to interannual.
J. Geophys. Res.
,
108
.
3251, doi:10.1029/2002JC001554
.
Kara
,
A. B.
,
P. A.
Rochford
, and
H. E.
Hurlburt
,
2002
:
Naval Research Laboratory mixed layer depth (NMLD) climatologies. NRL Rep. NRL/FR/7330-02-9995, 26 pp
.
Köhl
,
A.
,
R. H.
Käse
,
D.
Stammer
, and
N.
Serra
,
2007a
:
Causes of changes in the Denmark Strait overflow.
J. Phys. Oceanogr.
,
37
,
1678
1696
.
Köhl
,
A.
,
D.
Stammer
, and
B.
Cornuelle
,
2007b
:
Interannual to decadal changes in the ECCO global synthesis.
J. Phys. Oceanogr.
,
37
,
313
337
.
Large
,
W. G.
, and
S.
Pond
,
1981
:
Open ocean momentum flux measurements in moderate to strong winds.
J. Phys. Oceanogr.
,
11
,
324
336
.
Large
,
W. G.
, and
S.
Pond
,
1982
:
Sensible and latent heat flux measurements over the ocean.
J. Phys. Oceanogr.
,
12
,
464
482
.
Large
,
W. G.
,
J. C.
Williams
, and
S. C.
Doney
,
1994
:
Ocean vertical mixing: A review and a model with a nonlocal boundary layer parameterization.
Rev. Geophys.
,
32
,
363
403
.
Levitus
,
S.
, and
T. P.
Boyer
,
1994
:
Temperature. Vol. 4, World Ocean Atlas 1994, NOAA Atlas NESDIS 4, 117 pp
.
Levitus
,
S.
,
R.
Burgett
, and
T. P.
Boyer
,
1994
:
Salinity. Vol. 3, World Ocean Atlas 1994, NOAA Atlas NESDIS 3, 99 pp
.
Marshall
,
J.
,
A.
Adcroft
,
C.
Hill
,
L.
Perelman
, and
C.
Heisey
,
1997a
:
A finite-volume, incompressible Navier Stokes model for studies of the ocean on parallel computers.
J. Geophys. Res.
,
102
,
5753
5766
.
Marshall
,
J.
,
C.
Hill
,
L.
Perelman
, and
A.
Adcroft
,
1997b
:
Hydrostatic, quasi-hydrostatic, and nonhydrostatic ocean modeling.
J. Geophys. Res.
,
102
,
5733
5752
.
Melander
,
M. V.
,
N. J.
Zabusky
, and
J. C.
McWilliams
,
1988
:
Symmetric vortex merger in two dimensions: Causes and conditions.
J. Fluid Mech.
,
195
,
303
340
.
Nilsen
,
J. E. O.
, and
E.
Falck
,
2006
:
Variation of mixed layer properties in the Norwegian Sea for the period 1948–1999.
Prog. Oceanogr.
,
70
,
58
90
.
Nycander
,
J.
, and
J. H.
LaCasce
,
2004
:
Stable and unstable vortices attached to seamounts.
J. Fluid Mech.
,
507
,
71
94
.
Orvik
,
K. A.
,
2004
:
The deepening of the Atlantic water in the Lofoten Basin of the Norwegian Sea, demonstrated by using an active reduced gravity model.
Geophys. Res. Lett.
,
31
.
L01306, doi:10.1029/2003GL018687
.
Orvik
,
K. A.
, and
P.
Niiler
,
2002
:
Major pathways of Atlantic water in the northern North Atlantic and Nordic seas toward Arctic.
Geophys. Res. Lett
,
29
.
1896, doi:10.1029/2002GL015002
.
Pedersen
,
O. P.
,
M.
Zhou
,
K. S.
Tande
, and
A.
Evardsen
,
2005
:
Eddy formation on the coast of North Norway—Evidenced by synoptic sampling.
ICES J. Mar. Sci.
,
62
,
615
628
.
Poulain
,
P-M.
,
A.
Warn-Varnas
, and
P. P.
Niiler
,
1996
:
Near-surface circulation of the Nordic seas as measured by Lagrangian drifters.
J. Geophys. Res.
,
101
,
18237
18258
.
Rodionov
,
V. B.
,
1992
:
On the mesoscale structure of the frontal zones in the Nordic seas.
J. Mar. Syst.
,
3
,
127
139
.
Send
,
U.
, and
J.
Marshall
,
1995
:
Integral effects of deep convection.
J. Phys. Oceanogr.
,
25
,
855
872
.
Zhang
,
J. L.
, and
D.
Rothrock
,
2000
:
Modeling Arctic sea ice with an efficient plastic solution.
J. Geophys. Res.
,
105
,
3325
3338
.

Footnotes

Corresponding author address: Armin Köhl, Institut für Meereskunde, Zentrum für Meeres- und Klimaforschung, Universität Hamburg, Bundesstr. 53, 20146 Hamburg, Germany. Email: koehl@ifm.uni-hamburg.de