Elsevier

Progress in Oceanography

Volume 137, Part A, September 2015, Pages 149-172
Progress in Oceanography

Properties and pathways of Mediterranean water eddies in the Atlantic

https://doi.org/10.1016/j.pocean.2015.06.001Get rights and content

Highlights

  • Meddy characteristics are estimated as a function of meddy position.

  • The southern meddies are more compact, energetic and stable than the northern ones.

  • Within the first 600 km from the coast meddy merger is a common event.

  • Meddy decay is achieved within 3 years, but typically is 1–2 years due to topography.

  • The MAR and the Azores Current represent barriers for meddies.

Abstract

Data from ship vertical casts (NODC data-set), ARGO profiling floats (Coriolis data-set) and RAFOS-type neutral density floats (WOCE data-set) are used to study characteristics of meddies in the Northeast Atlantic. In total 241 Mediterranean water eddies (meddies) and 236 parts of float trajectories within meddies are selected for detailed analysis. The results suggest that the meddy generation rate at the southern and southwestern Iberian Peninsula (Portimao Canyon, cap St. Vincent, Estremadura Promontory, Gorringe Bank) is 3 times that at the northwestern Iberian Peninsula (Porto-Aveiro Canyons, Cape Finisterre and Galicia Bank).

Meddies generated south of Estremadura Promontory (the southern meddies), as compared to those generated north of it (the northern meddies), have smaller radii, smaller vertical extension, higher aspect ratio, higher Rossby number and higher stability (stronger potential vorticity anomaly). These latter properties result from the southern meddies higher relative vorticity and stronger buoyancy frequency anomaly.

Away from the generation regions, meddy drift concentrates along four main paths: three quasi-zonal paths (Northern, Central, Southern) and a path following the African coast (Coastal). The quasi-zonal paths are aligned to the isolines of the ambient potential vorticity field. Several cross-path exchanges, identified in this work, are aligned to topographic rises. Northward translation of the northern meddies within the North Atlantic Current to the subpolar gyre is detected.

Within the first 600 km from the coast, meddy merger is proved to be a common event. This explains the observed difference in radii between the newly generated meddies and those away from the Iberian margin.

The decay of the southern meddies proceeds mainly via the loss of their skirts and does not affect meddy cores until the latest stages. The decay of the northern meddies goes in parallel with the decay of their cores. In average meddy decay is achieved within 1–2 years, although may take over 3 years. Collisions with the Mid-Atlantic Ridge and seamounts sensibly decrease meddy lifetimes. Meddy decay also speeds up when meddies meet the Azores Current or the North Atlantic Current. A rapid drop in the number of meddies south of the Azores Current proves that it represents a dynamic barrier for weak meddies.

Introduction

Mediterranean water (MW) represents one of the major water masses in the Subtropical North Atlantic extending from the continental slope of Iberia to the Mid-Atlantic Ridge, and further on to the Western Atlantic. The outflow of the MW through the Gibraltar Strait presents exceptionally strong anomalies of temperature, salinity, nutrients, sediments, etc. for Atlantic mid-depths, which makes it unique in the World Ocean. It largely influences the physical and chemical properties, as well as the stratification of the water column from the sea-surface to, at least, 2000 m depth (Mauritzen et al., 2001). Studies of the mechanisms of the MW spreading in the Atlantic bring up the two main hypothesis (Bower et al., 1997, Mazé et al., 1997, Iorga and Lozier, 1999a, Iorga and Lozier, 1999b, Sparrow et al., 2002, Lozier and Stewart, 2008): the advective transport and the integrated transport by Mediterranean water eddies.

Mediterranean water eddies (meddies) are generated from instabilities of the Mediterranean Undercurrent (MUC) and are encountered across the North-East Atlantic basin from Iberia to the Mid-Atlantic ridge and from tropical to mid-latitudes (Richardson et al., 2000). Meddies are mesoscale vortices, 10–50 km in radius, with one or two vertically aligned cores typically centered between 800 and 1200 m and detected in temperature-salinity sections as strong lens-like anomalies in temperature (up to 4 °C) and salinity (up to 1) (Richardson et al., 2000). Meddies rotate anticyclonically with azimuthal velocities up to 50 cm s−1 (Richardson et al., 2000). Since the MUC water is rather homogeneous as compared to the North Atlantic Central Water, meddy cores are characterized by a strong potential vorticity anomaly, further enhanced by the anticyclonic rotation of their cores. The gradient of potential vorticity at a meddy boundary presents a barrier for particle exchange with the ambient fluid, which is a reason for long lifetimes of meddies.

A single meddy transports 109–1011 tons of salt (Shapiro et al., 1995b, L’Hégaret et al., 2014) hundreds of kilometers away from the Iberian coast, with an average velocity of 1–5 cm s−1, releasing the salt and heat to the ocean. The estimates of the role of meddies (and MW cyclones) in the formation of the Mediterranean salt tongue in the Atlantic vary widely. Bower et al. (1997), based on one year of observations, calculated that 15–20 meddies generated per year, is enough to support about 50% of MW salt flux. This corresponds to the estimate of 17 meddies generated per year by Richardson et al. (2000), as well as to the independent estimate of the meddy salt flux by Arhan et al. (1994). Meanwhile, some of the formation regions, like Portimao Canyon and the Gorringe Bank, were not taken into account (Serra and Ambar, 2002) and the meddy flux is possibly underestimated. For example, in the model study by Barbosa Aguiar et al. (2013) the annual meddy formation rate has been evaluated to reach 26 meddies per year. Mazé et al. (1997), using 3 hydrographical surveys in the Iberian basin, concluded that 100% of the westward MW spreading is due to meddies. Sparrow et al. (2002), using trajectories of RAFOS drifters, inferred that the MW advection plays an important role north of 36°N, while south of this latitude, the transport is mainly due to meddies. On the other hand, Barbero et al. (2010) did not find any persistent advective MW transport in the region 39–45°N and 16–21°W.

As mentioned above, Meddies are formed from the MUC. The main formation regions are: Portimão Canyon (Serra and Ambar, 2002), Cape St. Vincent (Prater and Sanford, 1993, Richardson et al., 2000), Lisbon and Setubal Canyons just south of Estremadura Promontory and Nazare Canyon just north of it (Richardson et al., 2000). A possible meddy generation at the Gorringe Bank (Serra and Ambar, 2002), Porto and Aveiro Canyons (Chérubin et al., 1997) has been discussed in literature. In this paper we argue that meddies may also form at Cape Finisterre and Galicia Bank. Meddies might also form north or northeast of the Iberian Peninsula. Thus, a meddy observed just north of the Iberian Peninsula is thought to be formed in the Gulf of Biscay (Carton et al., 2013). Paillet et al. (2002) suggested that their meddy Ulla could have been formed near Cape Ortegal (at the northern coast of the Iberian Peninsula).

Two main mechanisms of meddy formation have been proposed: the baroclinic instability of the MUC or of its seawards branches (McWilliams, 1985, Chérubin et al., 2007) and the detachment of the shear lateral/bottom boundary layer at capes and canyons (D’Asaro, 1988, Aiki and Yamagata, 2004). Computations suggest that already 100 km downstream of the Gibraltar Strait (i.e. east of Portimão Canyon) bottom friction on the continental slope can generate a relative vorticity in the MUC, similar to that observed in the newly formed meddies: that is 0.2–0.3f, where f is the Coriolis parameter (Prater, 1992).

In their model study, Aiki and Yamagata (2004) also showed that a rapid change in direction of a coastline (i.e. a cape or a promontory) may not be indispensable for the early generation of a meddy, but is absolutely necessary for meddies to separate from the continental shelf. Therefore, meddy detached at capes may be generated upstream (relative to the direction of the MUC). As a meddy detach from the continental slope, a surface intensified cyclone is often formed upstream (Richardson et al., 2000, Serra et al., 2002, Aiki and Yamagata, 2004, L’Hégaret et al., 2014). Being intensified, the cyclone advects a meddy around its boundary, forcing it away from the continental slope (Richardson et al., 2000). Aiki and Yamagata (2004) relate formation of such cyclones with the instability of the surface upwelling current at the capes. Such cyclones may also be formed as a consequence of the upper ocean potential vorticity conservation at the lee side of a drifting meddy (Bashmachnikov and Carton, 2012), or, for of the MUC instabilities in canyons, as a part of coupled mushroom-like structures (Serra et al., 2005).

After detachment from the continental slope, meddies propagate westwards. Four major meddy pathways have been identified in regional studies (Shapiro and Meschanov, 1996, Demidov et al., 2012): from the Portimao-St. Vincent region to the Canary Islands; from St. Vincent-Estremadura region along the southern flank of Josephine seamount and farther southwest; from St. Vincent-Estremadura region along the northern flank of Josephine seamount and farther west; and northwestward along the southern flank of the Galicia Bank (Demidov et al., 2012). Demidov et al. (2012) mention that about 70% of meddies leave the Iberian Peninsular to travel west-southwest and only a third of them drifts north or northeast.

For mesoscale eddies their sufficiently large horizontal scales allow the planetary β-effect to efficiently affect their propagation (Nof, 1983). One mechanism of meddy self-translation is the generation of secondary circulations in its core (β-gyres) in a spatially varying background potential vorticity field (Cushman-Roisin et al., 1990). After a meddy has started its motion, the surrounding fluid is forced sidewise, and the same background potential vorticity gradient generates the ambient vorticity torques, which enhance the translation. As a result, a meddy self-translates along the isolines of the background potential vorticity, on the planetary β-plain – to the west, with a characteristics velocity Vm-βRd2(1+ΔH/Ha) (Marshall, 1988, Cushman-Roisin et al., 1990). Here Rd is the Rossby radius of deformation taken as the measure of the eddy horizontal scale, Ha is the mean thickness of the density layer where the eddy resides and ΔH is the isopycnal deviation at the eddy center (Cushman-Roisin et al., 1990, Morrow et al., 2004).

The entrainment of the surrounding fluid in the gradient of the background potential vorticity also results in a southward drift of a meddy, which is at least an order of magnitude smaller than its westward self-translation (Cushman-Roisin et al., 1990). This southwards drift can also result from the dissipation of a meddy via lateral intrusions or via the radiation of Rossby waves (Flierl, 1984, Colin de Verdière, 1992): the consequent decrease of the relative vorticity in the meddy core induces the drift by virtue of potential vorticity conservation in the core.

The dynamic similarity of the topographic β-effect or the baroclinic β-effect to the planetary β-effect for quasi-geostrophic motions, means that the eddy propagation direction/velocity depends on the relative intensity of the β-effects and on the direction of the corresponding potential vorticity gradients. The intensity of the β-gyres, generated by the baroclinic β-effect of the mean flow, increases with the increase of the intensity of the flow (Morel and McWilliams, 1997). This self-translation mechanism is directed against eddy advection by a zonal flow and can move mesoscale eddies (including meddies) against the dominant flow (Morel and McWilliams, 1997). This counterflow propagation holds for a zonal current invariant with depth (van Leeuwen, 2007), as well as for a vertically sheared one (Vandermeirsch et al., 2001). It is not observed for submesoscale vortices, for which the pure advection by background current dominates (Dewar and Meng, 1995). On another hand, in a southward background current the baroclinic β-effect is directed with the mean flow, and this flow can be efficient in advection of meddies (Morel, 1995).

Additionally, an anticyclonic vortex in a large-scale shear background flow tends to migrate towards the current axis in the area of opposite background vorticity and away from the current axis in the area of the like-signed background vorticity (Bell, 1990). The similar behavior is obtained for a meddy interacting with a jet-like flow, the horizontal scale of which is comparable with the meddy diameter (Vandermeirsch et al., 2003). In particular, for a meddy approaching from north the eastwards Azores Current, this means crossing the jet. The mechanism of crossing the jet lies in generation of an anticyclonic meander upstream and a cyclonic meander downstream in the jet. The cyclone, being intensified, pushes the meddy south. But only a sufficiently strong meddy can cross a jet, i.e. if its maximal azimuthal velocity exceeds that in the axis of the jet (Vandermeirsch et al., 2003). After crossing, the meddy may be expelled to the south, or stay trapped at the southern boundary of the jet. This depends on the relative intensity of the cyclone and the anticyclone simultaneously formed in the jet by the meddy (Vandermeirsch et al., 2003). For a wide background flow, a sufficient condition for trapping at its southern limit is shielding the meddy at the south with an area of enhanced ambient negative relative vorticity (Bell, 1990). The analysis of meddy trajectories (Richardson and Tychensky, 1998, Tychensky and Carton, 1998, Bashmachnikov et al., 2009b) demonstrate a fast meddy translation across the Azores Current (meddies Hyperion and Ceres), as well as a meddy trapping, for at least 6 months, at the southern boundary of the Azores Current (meddy Encelade).

When coupled with a cyclone of the same strength in the same layer, a meddy also tends to drift along isolines of the background potential vorticity (Velasco Fuentes and van Heijst, 1995). Lager and stronger anticyclones or cyclones (including the surface ones) are seen by a weaker meddy as strong local anomalies of potential vorticity. The former are found to sharply change the direction of meddy propagation, advecting the latter around their periphery (Schultz Tokos et al., 1994, Richardson et al., 2000, Carton et al., 2010).

Meddies, separated from the surrounding waters by strong gradients of potential vorticity, are quite stable structures with life times often exceeding 1 year. The mechanisms of meddy decay can include gradual vertical and horizontal diffusive exchange with the background due to instability of meddy periphery (Hua et al., 2013), along-isopycnal lateral intrusions (Hebert et al., 1990), entrainment of meddy periphery by a background flow (Mariotti et al., 1994), internal meddy instability resulting in core deformation and in filaments formation (Ménesguen et al., 2012), energy dispersion into lee Rossby waves (Flierl, 1984), loss of a part of meddy core, splitting or complete decay while interacting with seamounts (Richardson et al., 2000, Cenedese, 2002, Bashmachnikov et al., 2009a, Sokolovskiy et al., 2013).

Hebert et al. (1990), in their 2 year-long periodic re-sampling of meddy Sharon, conclude that lateral intrusions are a more efficient mechanism of meddy decay than horizontal/vertical diffusion. The former mechanism, though, cannot compete in its destructive efficiency with the more rare, but more drastic meddy-seamount interaction. The Meddy seamount interaction, in turn, can be strong and lead to meddy significant or complete erosion, moderate in the case of meddy splitting and possible reunification behind the seamount, or weak in the case of meddy rotation around a seamount flank, leaving the seamount without any significant change of its properties (Shapiro et al., 1995a, Richardson et al., 2000, Cenedese, 2002, Herbette et al., 2003, Adduce and Cenedese, 2004, Sokolovskiy et al., 2013). The strength of the interaction depends on horizontal seamount dimensions in comparison to the radius of the meddy, as well as on the height of the seamount (van Geffen and Davies, 2000, Sokolovskiy et al., 2013) and is enhanced in the presence of a background flow (Cenedese, 2002). In-situ observations and numerical models suggest that during interactions with a seamount a meddy typically loses around 25–40% of its core (Shapiro et al., 1995a, Richardson et al., 2000, Wang and Dewar, 2003, Bashmachnikov et al., 2009a).

In spite of a significant progress in our understanding of meddy pathways and dynamics in the recent decades, the observational evidences of the theoretical and model results mostly cover case studies of some parts of meddy life cycles (Shapiro et al., 1995b, Richardson et al., 2000, Demidov et al., 2012). This work presents the most complete for the moment, systematic and holistic overview of observations of meddies, and gives further details on meddy characteristics and dynamics, as a function of their distance from the generation region and latitude.

Section snippets

Hydrological data

Data from the World Ocean Database 2013 (WOD) were downloaded from the National Oceanographic Data Center (NODC, http://www.nodc.noaa.gov/OC5/WOD/pr_wod.html): OSD (Ocean Station Data, low vertical resolution), CTD (Conductivity–Temperature–Depth, high vertical resolution) and PFL (Profiling Floats, with various vertical resolutions, mainly obtained from the ARGO profiling float array). We only use data collected between 1950 and 2012. The climatological reference state is obtained from the

Meddy distribution and pathways

In this section, the probability to find a meddy in different parts of the Atlantic is computed as a ratio of the number of casts through meddies to the total number of casts in 1° × 1° areas. To avoid local biases in the probability distribution, resulting from in-purpose multiple sampling of the same meddy during a cruise, the data-points, sampled in approximately the same place and approximately at the same time, are clustered to a single point. The clustering is done separately for meddy and

Conclusions and discussion

In this study we combined data from ship vertical casts (NODC data-set), ARGO profiling floats (Coriolis data-set) and RAFOS-type neutral density floats (WOCE data-set) to derive of spatial variations of meddy characteristics in the NE Atlantic. From the overall 26,062 vertical casts available, 775 meddies were identified using Richardson’s criterion (salinity anomaly exceeds 0.2 in the MW layer, Richardson et al., 1991). 241 of the meddies, sufficiently covered with observations, were selected

Acknowledgements

The authors acknowledge the scientific project MEDTRANS (PTDC/MAR/117265/2010), sponsored by the Portuguese Foundation for Science and Technology (FCT) and Marine and Environmental Sciences Center (MARE) of the University of Lisbon (CO-Pest-OE/MAR/UI0199/2011). I.B. also acknowledges the contract C2008-UL-CO-3 of Ciencia 2008 between Foundation for Science and Technology (FCT) and the University of Lisbon (UL). The authors acknowledge Joana Medeiros for collecting meddy characteristics from

References (96)

  • J.P. Mazé et al.

    Volume budget of the eastern boundary layer off the Iberian Peninsula

    Deep-Sea Research I

    (1997)
  • R.D. Pingree et al.

    Structure of a Meddy (Bobby 92) southeast of the Azores

    Deep-Sea Research

    (1993)
  • M. Rhein et al.

    Modification of Mediterranean water in the Gulf of Cadiz, studied with hydrographic, nutrient and chlorofluoromethane data

    Deep-Sea Research I

    (1993)
  • P.L. Richardson et al.

    A search for meddies in historical data

    Dynamics of Atmospheres and Oceans

    (1991)
  • P.L. Richardson et al.

    A census of meddies tracked by floats

    Progress in Oceanography

    (2000)
  • N. Serra et al.

    Eddy generation in the Mediterranean undercurrent

    Deep-Sea Research I

    (2002)
  • N. Serra et al.

    Observations and numerical modelling of the Mediterranean outflow splitting and eddy generation

    Deep-Sea Research II

    (2005)
  • G.I. Shapiro et al.

    Spreading pattern and mesoscale structure of Mediterranean outflow in the Iberian Basin estimated from historical data

    Journal of Marine Systems

    (1996)
  • J.H.G.M. van Geffen et al.

    A monopolar vortex encounters an isolated topographic feature on a β-plane

    Dynamics of Atmospheres and Oceans

    (2000)
  • F.O. Vandermeirsch et al.

    Interaction between an eddy and a zonal jet: Part II. Two-and-a-half-layer model

    Dynamics of Atmospheres and Oceans

    (2003)
  • C. Adduce et al.

    An experimental study of a mesoscale vortex colliding with topography of varying geometry in a rotating fluid

    Journal of Marine Research

    (2004)
  • H. Aiki et al.

    A numerical study on the successive formation of meddy-like lenses

    Journal of Geophysical Research

    (2004)
  • I. Ambar et al.

    A physical and chemical description of the Mediterranean outflow in the Gulf of Cadiz

    Deutsche Hydrographische Zeitschrift

    (1976)
  • M. Arhan et al.

    The eastern boundary of the Subtropical North Atlantic

    Journal of Physical Oceanography

    (1994)
  • L. Armi et al.

    Large lenses of highly saline Mediterranean water

    Journal of Physical Oceanography

    (1984)
  • L. Armi et al.

    Two years in the life of a Mediterranean salt lens

    Journal of Physical Oceanography

    (1989)
  • R.R. Bambrey et al.

    Strong interactions between two corotating quasi-geostrophic vortices

    Journal of Fluid Mechanics

    (2007)
  • L. Barbero et al.

    Variability of the water mass transports and fluxes in the eastern North Atlantic during 2001

    Journal of Geophysical Research

    (2010)
  • M.O. Baringer et al.

    Mixing and spreading of the Mediterranean outflow

    Journal of Physical Oceanography

    (1997)
  • I. Bashmachnikov et al.

    Surface signature of Mediterranean water eddies in the Northeastern Atlantic: effect of the upper ocean stratification

    Ocean Science

    (2012)
  • I. Bashmachnikov et al.

    In-situ and remote sensing signature of meddies east of the Mid-Atlantic ridge

    Journal of Geophysical Research

    (2009)
  • I. Bashmachnikov et al.

    Characteristics of surface signatures of Mediterranean water eddies

    Journal of Geophysical Research

    (2014)
  • I. Bashmachnikov et al.

    Temperature–salinity distribution in the Northeast Atlantic from ship and Argo vertical casts

    Ocean Science

    (2015)
  • G.I. Bell

    Interaction between vortices and waves in a simple model of geophysical flow

    Physics of Fluids

    (1990)
  • A.S. Bower et al.

    Lagrangian observations of Meddy formation during a Mediterranean undercurrent seeding experiment

    Journal of Physical Oceanography

    (1997)
  • A. Bower et al.

    Structure of the Mediterranean undercurrent and Mediterranean water spreading around the southwestern Iberian Peninsula

    Journal of Geophysical Research

    (2002)
  • X. Carton

    Hydrodynamical modeling of oceanic vortices

    Surveys In Geophysics

    (2001)
  • X. Carton et al.

    The generation of tripoles from unstable axisymmetric isolated vortex structures

    Europhysics Letters

    (1989)
  • X. Carton et al.

    Meddy dynamics and interaction with neighboring eddies southwest of Portugal: observations and modeling

    Journal of Geophysical Research

    (2010)
  • C. Cenedese

    Laboratory experiments on mesoscale vortices colliding with a seamount

    Journal of Geophysical Research

    (2002)
  • D.B. Chelton et al.

    Geographical variability of the first baroclinic Rossby radius of deformation

    Journal of Physical Oceanography

    (1998)
  • L. Chérubin et al.

    Descriptive analysis of the hydrology and currents on the Iberian shelf from Gibraltar to Cape Finisterre: preliminary results from the SEMANE and INTERAFOS experiments

    Annales Hydrographiques

    (1997)
  • L.M. Chérubin et al.

    Vortex dipole formation by baroclinic instability of boundary currents

    Journal of Physical Oceanography

    (2007)
  • A. Colin de Verdière

    On the southward motion of Mediterranean salt lenses

    Journal of Physical Oceanography

    (1992)
  • B. Cushman-Roisin

    Environmental Fluid Mechanics

    (2010)
  • B. Cushman-Roisin et al.

    Westward motion of mesoscale eddies

    Journal of Physical Oceanography

    (1990)
  • E.A. D’Asaro

    Generation of submesoscale vortices: a new mechanism

    Journal of Geophysical Research

    (1988)
  • A.N. Demidov et al.

    Detection of Mediterranean lenses in the Atlantic Ocean by profilers of the Argo Project

    Oceanology

    (2012)
  • Cited by (48)

    • Recent changes in the Mediterranean Sea

      2022, Oceanography of the Mediterranean Sea: An Introductory Guide
    View all citing articles on Scopus
    View full text