• Allen, J., , D. Smeed, , and A. Chadwick, 1994: Eddies and mixing at the Iceland–Faroes Front. Deep-Sea Res., 41, 5179.

  • Beaird, N. L., , I. Fer, , P. B. Rhines, , and C. C. Eriksen, 2012: Dissipation of turbulent kinetic energy inferred from Seagliders: An application to the eastern Nordic seas overflows. J. Phys. Oceanogr., 42, 22682282.

    • Search Google Scholar
    • Export Citation
  • Bretherton, F. P., , R. E. Davis, , and C. B. Fandry, 1976: A technique for objective analysis and design of oceanographic experiments applied to MODE-73. Deep-Sea Res., 23, 559582.

    • Search Google Scholar
    • Export Citation
  • Cenedese, C., , J. A. Whitehead, , T. A. Ascarelli, , and M. Ohiwa, 2004: A dense current flowing down a sloping bottom in a rotating fluid. J. Phys. Oceanogr., 34, 188203.

    • Search Google Scholar
    • Export Citation
  • Darelius, E., , I. Fer, , and D. Quadfasel, 2011: Faroe Bank Channel overflow: Mesoscale variability. J. Phys. Oceanogr., 41, 21372154.

  • Duncan, L. M., , H. L. Bryden, , and S. A. Cunningham, 2003: Friction and mixing in the Faroe Bank Channel outflow. Oceanol. Acta, 26, 473486.

    • Search Google Scholar
    • Export Citation
  • Enmar, L., , K. Borenäs, , I. Lake, , and P. Lundberg, 2009: Comments on “Is the Faroe Bank Channel overflow hydraulically controlled?” J. Phys. Oceanogr., 39, 15341538.

    • Search Google Scholar
    • Export Citation
  • Eriksen, C. C., , T. J. Osse, , R. D. Light, , T. Wen, , T. W. Lehman, , P. L. Sabin, , J. W. Ballard, , and A. M. Chiodi, 2001: Seaglider: A long-range autonomous underwater vehicle for oceanographic research. IEEE J. Oceanic Eng., 26, 424436.

    • Search Google Scholar
    • Export Citation
  • Fer, I., , G. Voet, , K. S. Seim, , B. Rudels, , and K. Latarius, 2010: Intense mixing of the Faroe Bank Channel overflow. Geophys. Res. Lett., 37, L02604, doi:10.1029/2009GL041924.

    • Search Google Scholar
    • Export Citation
  • Fogelqvist, E., , J. Blindheim, , T. Tanhua, , S. Østerhus, , E. Buch, , and F. Rey, 2003: Greenland–Scotland overflow studied by hydro-chemical multivariate analysis. Deep-Sea Res. I, 50, 73102.

    • Search Google Scholar
    • Export Citation
  • Geyer, F., , S. Østerhus, , B. Hansen, , and D. Quadfasel, 2006: Observations of highly regular oscillations in the overflow plume downstream of the Faroe Bank Channel. J. Geophys. Res., 111, C12020, doi:10.1029/2006JC003693.

    • Search Google Scholar
    • Export Citation
  • Gill, A. E., 1982: Atmosphere–Ocean Dynamics. Academic Press, 662 pp.

  • Girton, J. B., , L. J. Pratt, , D. A. Sutherland, , and J. F. Price, 2006: Is the Faroe Bank Channel overflow hydraulically controlled? J. Phys. Oceanogr., 36, 23402349.

    • Search Google Scholar
    • Export Citation
  • Hansen, B., , and J. Meincke, 1979: Eddies and meanders in the Iceland–Faroe Ridge area. Deep-Sea Res., 26, 10671082.

  • Hansen, B., , and S. Østerhus, 2000: North Atlantic–Nordic seas exchanges. Prog. Oceanogr., 45, 109–208.

  • Hansen, B., , and S. Østerhus, 2007: Faroe Bank Channel overflow 1995–2005. Prog. Oceanogr., 75, 817856.

  • Harvey, J. G., , and A. Theodorou, 1986: The circulation of Norwegian Sea overflow water in the eastern North Atlantic. Oceanol. Acta, 9, 393402.

    • Search Google Scholar
    • Export Citation
  • Hermann, F., 1967: The T-S diagram analysis of the water masses over the Iceland–Faroe Ridge and in the Faroe Bank Channel (Overflow ’60). Rapp. P.-V. Reun. - Cons. Int. Explor. Mer,157, 139–149.

  • Johnson, G. C., 2008: Quantifying Antarctic Bottom Water and North Atlantic Deep Water volumes. J. Geophys. Res.,113, C05027, doi:10.1029/2007JC004477.

  • Johnson, G. C., , and T. B. Sanford, 1992: Secondary circulation in the Faroe Bank Channel outflow. J. Phys. Oceanogr., 22, 927933.

  • Killworth, P., 2001: On the rate of descent of overflows. J. Geophys. Res., 106 (C10), 22 26722 275.

  • Mauritzen, C., , J. Price, , T. Sanford, , and D. Torres, 2005: Circulation and mixing in the Faroese Channels. Deep-Sea Res., 52, 883913.

  • Meincke, J., 1972: The hydrographic section along the Iceland–Faroe Ridge carried out by R.V. ‘Anton Dohrn’ in 1959–1971. Ber. Dtsch. Wissenchaftlichen Komm. Meeresforschung, 22, 372384.

    • Search Google Scholar
    • Export Citation
  • Nof, D., , N. Paldor, , and S. V. Gorder, 2002: The reddy maker. Deep-Sea Res., 49, 15311549.

  • Perkins, H., , T. J. Sherwin, , and T. Hopkins, 1994: Amplification of tidal currents by overflow on the Iceland–Faroe Ridge. J. Phys. Oceanogr., 24, 721735.

    • Search Google Scholar
    • Export Citation
  • Perkins, H., , T. S. Hopkins, , S. A. Malmberg, , P. M. Poulain, , and A. Warn-Varnas, 1998: Oceanographic conditions east of Iceland. J. Geophys. Res., 103 (C10), 21 53121 542.

    • Search Google Scholar
    • Export Citation
  • Prater, M. D., , and T. Rossby, 2005: Observations of the Faroe Bank Channel outflow using bottom-following RAFOS floats. Deep-Sea Res., 52, 481494.

    • Search Google Scholar
    • Export Citation
  • Pratt, L. J., , U. Riemenschneider, , and K. R. Helfrich, 2007: A transverse hydraulic jump in a model of the Faroe Bank Channel outflow. Ocean Modell., 19, 19.

    • Search Google Scholar
    • Export Citation
  • Price, J. F., , and M. O. Baringer, 1994: Overflows and deep water production by marginal seas. Prog. Oceanogr., 33, 161200.

  • Riemenschneider, U., , and S. Legg, 2007: Regional simulations of the Faroe Bank Channel overflow in a level model. Ocean Modell., 17, 93122.

    • Search Google Scholar
    • Export Citation
  • Saunders, P. M., 1990: Cold outflow from the Faroe Bank Channel. J. Phys. Oceanogr., 20, 2943.

  • Saunders, P. M., 1996: The flux of dense cold overflow water southeast of Iceland. J. Phys. Oceanogr., 26, 8595.

  • Seim, K. S., , and I. Fer, 2011: Mixing in the stratified interface of the Faroe Bank Channel overflow: The role of transverse circulation and internal waves. J. Geophys. Res., 116, C07022, doi:10.1029/2010JC006805.

    • Search Google Scholar
    • Export Citation
  • Seim, K. S., , I. Fer, , and J. Berntsen, 2010: Regional simulations of the Faroe Bank Channel overflow using a σ-coordinate ocean model. Ocean Modell., 35, 3144.

    • Search Google Scholar
    • Export Citation
  • Steele, J. H., 1959: Observations of deep water overflow across the Iceland–Faroe Ridge. Deep-Sea Res., 6, 7072.

  • Swift, J., 1984: The circulation of the Denmark Strait and Iceland–Scotland overflow waters in the North Atlantic. Deep-Sea Res., 31, 13391355.

    • Search Google Scholar
    • Export Citation
  • Wilkenskjeld, S., , and D. Quadfasel, 2005: Response of the Greenland–Scotland overflow to changing deep water supply from the Arctic Mediterranean. Geophys. Res. Lett., 32, L21607, doi:10.1029/2005GL024140.

    • Search Google Scholar
    • Export Citation
  • Xu, X., , W. Schmitz, , H. Hurlburt, , P. Hogan, , and E. Chassignet, 2010: Transport of Nordic seas overflow water into and within the Irminger Sea: An eddy-resolving simulation and observations. Geophys. Res. Lett., 115, C12048, doi:10.1029/2010JC006351.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (top) Bathymetry and location of features in the area of study. Blue lines indicate the locations of synthetic sections discussed later in the text. Numbered colored squares mark the locations of the profiles in Fig. 3. Thick contours are plotted at 500-m intervals beginning with the 500-m isobath and continuing deeper. (bottom) A profile of ridge crest depth along the dashed line in the top panel with Western Valley (WV), IFR, FBC, and Faroe Bank (FB) labeled.

  • View in gallery

    (top) Salinity contours on a section normal to the IFR occupied by an SG on 3–23 Dec 2007. Dive and climb positions are shown as black vertical tick marks along the upper x axis. Location of the track is shown in the inset map. (bottom) Absolute geostrophic velocity [cross-section component, color contours (m s−1)] and potential density [black contours (kg m−3)]. Positive velocities are toward the northwest.

  • View in gallery

    Potential temperature vs salinity for six profiles at the location of the numbered colored squares in Fig. 1. Water mass definitions for NAW, MNAW, MEIW, NSAIW, NSDW, and ISOW are shown as gray boxes covering the property ranges of each type. Potential density is contoured in gray.

  • View in gallery

    Bottom temperatures (°C) on the Iceland–Faroe Ridge from all SG dives between November 2006 and 2009. Shaded bathymetric contours are plotted every 250 m, with thick contours every 500 m.

  • View in gallery

    (a) Location of section I in FBC and averaged DAC vectors (red) used to reference geostrophic velocity calculations. Depth contours at 100-m intervals with 500-m contour shown by thick line. Depth (m) by cross-channel distance (km) sections of (b) potential temperature (°C) and (c) absolute geostrophic velocity (m s−1; positive to the northwest). The mean depth of the 3°C isotherm observed in individual dives is plotted in black with dashed lines indicating plus or minus one std dev.

  • View in gallery

    (top) Thickness (m) of the ISOW layer near the FBC exit based on a fine-grid objective map of all SG data. Plume thickness is the vertical distance between the mapped σθ = 27.8 kg m−3 isopycnal surface and the seafloor. Bathymetric contours at 50-m intervals with 1000- and 500-m isobaths shown by thick lines. The deepest point is indicated by a black-and-white dashed line. The portion of the 1000-m isobath where the downslope plume edge is deeper than the reach of the SG is highlighted in pink, indicating the area through which the deeper branch of the plume flows. The primary and terminal (second) sills of the FBC are labeled, along with the topographic bump, which marks the end of the plume bifurcation region. (bottom) Near-bottom temperatures (°C) from SG dives in the FBC outflow region. The bathymetry is the same as the top panel.

  • View in gallery

    (a) Mean overflow potential density contours and (b) observations of the depth of the 4.5°C isotherm for each of the 13 composite sections shown in Fig. 1. Sections are offset by 300-m depth for clarity, with vertical tick marks placed every 100 m. Labeled depths refer to section I. Section roman numerals are centered at 600 m in the reference frame of each section. Dark lines indicate bathymetry along each section. Width is measured with respect to the location of the 700-m isobath. In (b), black dots show observations of the depth of the 4.5°C isotherm while the thin gray line shows the mean depth.

  • View in gallery

    Cross-sectional transport through composite sections I–XIII (SG composite) vs distance from FBC sill. Std dev from the Monte Carlo approach are shown by the pink patch. The average transport through individual SG sections in the WV is shown as well. Included are transport estimates from various authors: blue symbols indicate estimates using the σθ ≥ 27.65 kg m−3 criterion, whereas red symbols use σθ ≥ 27.8 kg m−3. Black markers indicate other plume definitions used by various authors [model tracers in Riemenschneider and Legg (2007), depth of no motion in Perkins et al. (1998), and waters colder than 3°C in Darelius et al. (2011)].

  • View in gallery

    (left) Mean bottom potential temperature (°C) in 7.5 km × 7.5 km boxes. (right) Min bottom potential temperature anomaly from the mean in the same boxes (°C). Numbered and outlined boxes in the left panel correspond to the profiles in Fig. 10.

  • View in gallery

    (top) All potential temperature profiles and (bottom) all Θ–S curves contained in the four boxes outlined in black in Fig. 9. From the FBC box on the right (box 1) to the left, the boxes move downstream. The thick contour on Θ–S plots is the 27.8 kg m−3 isopycnal with 0.1 kg m−3 intervals. Blue profiles show upstream conditions in the Nordic seas and red profiles show ambient conditions in the Iceland Basin.

  • View in gallery

    Bottom-layer property means (solid line) and std dev (dashed line) on the IFR between the 500- and 1000-m isobaths in 15-km bins of distance from the primary FBC sill. Individual observations are plotted in light gray. (top) BML temperature. (bottom) HAB of the 4.5°C isotherm. The vertical gray lines around 50 and 80 km indicate the location of the secondary FBC sill and the point at which the FBC overflow first intersects the 1000-m isobath, respectively.

  • View in gallery

    Near-bottom absolute geostrophic velocities (75 m above bottom) and RAFOS float trajectories (Prater and Rossby 2005). Vector color indicates potential density at 75 m above bottom.

  • View in gallery

    Distribution of variables related to eddy and mean baroclinic velocities on the AIFR. Thick black lines represent the x-ridge group and medium gray lines represent the a-ridge group in (a)–(e). In (c), the light gray line represents velocities based on topographic slopes. (a) Distribution of vertically averaged plume density anomaly from ambient Iceland Basin density. (b) Isopycnal slopes between dive pairs. (c) Magnitude of the velocity jump across the plume interface due to the isopycnal slope and density anomaly as in Eq. (3). The light gray line shows the velocity jump if Sisp is replaced by the local topographic gradient. (d) Distribution of the along-section BML temperature gradient between sequential dives. (e) Distribution of the ratio of isopycnal to topographic slope. (f) Distribution of the ratio of velocity jumps due to eddy isopycnal slopes to that due to topographic slope for the x-ridge group (thin black line), a-ridge group (gray line), and all dives (thick black line).

  • View in gallery

    Bottom temperatures from individual dives in the Western Valley showing anomalous cold waters along the Iceland shelf.

  • View in gallery

    Trajectories of particles arriving at section XIII (Fig. 1) computed using the constant descent rate (1/600) model of Killworth (2001). The model is run backward for 300 km from the blue points along section XIII. Color is an index of the position along section XIII of each arriving trajectory.

  • View in gallery

    Four representative SG sections intersecting the Iceland shelf in September 2007 and in June 2007, 2008, and 2009. Contours of σΘ are shown in color for the portion of the sections with density greater than 27.65 kg m−3. Two of the sections encounter large overflow events and two see little overflow.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 211 211 24
PDF Downloads 155 155 10

Overflow Waters at the Iceland–Faroe Ridge Observed in Multiyear Seaglider Surveys

View More View Less
  • 1 School of Oceanography, University of Washington, Seattle, Washington
© Get Permissions
Full access

Abstract

This paper presents new observations of the overflow waters downstream of the Faroe Bank Channel (FBC) and the Iceland–Faroe Ridge (IFR). Between 2006 and 2009, over 17 400 hydrographic profiles were collected during quarterly deployments in the region by autonomous gliders, providing previously unrealized spatial resolution to observations downstream of the FBC. Observations show that the second sill of the FBC coincides with the largest changes in the overflow plume, including significant thinning, widening, and entrainment. Between the second sill and a topographic feature 75 km downstream, the plume bifurcates with the densest portion (65% of the transport), descending below 1000 m. On the IFR, near-bottom velocities are directed alongslope with speeds averaging 21.5 cm s−1. Observations indicate that 80% of baroclinic velocities associated with mesoscale variability of the overflow plume are smaller than the alongslope topographically induced circulation. Evidence of overflow is found at all locations on the Atlantic flank of the IFR. However, the meridionally oriented portion at 13°W has anomalously warm bottom water and divides FBC and eastern IFR overflow from overflow found in the Western Valley. Individual Seaglider sections identify IFR overflow in a narrow current (8–14 km wide) along the Iceland shelf with a mean transport of 0.43 Sv (1 Sv ≡ 106 m3 s−1) with significant variability from days to weeks. A lower-bound estimate of 0.8 Sv of total IFR overflow is presented. These results provide constraints on regional models that inform the representation of this crucial, yet underresolved, region in large-scale ocean and climate models.

Corresponding author address: Nick Beaird, School of Oceanography, University of Washington, 1492 Boat St., Seattle, WA 98195. E-mail: nlbeaird@uw.edu

Abstract

This paper presents new observations of the overflow waters downstream of the Faroe Bank Channel (FBC) and the Iceland–Faroe Ridge (IFR). Between 2006 and 2009, over 17 400 hydrographic profiles were collected during quarterly deployments in the region by autonomous gliders, providing previously unrealized spatial resolution to observations downstream of the FBC. Observations show that the second sill of the FBC coincides with the largest changes in the overflow plume, including significant thinning, widening, and entrainment. Between the second sill and a topographic feature 75 km downstream, the plume bifurcates with the densest portion (65% of the transport), descending below 1000 m. On the IFR, near-bottom velocities are directed alongslope with speeds averaging 21.5 cm s−1. Observations indicate that 80% of baroclinic velocities associated with mesoscale variability of the overflow plume are smaller than the alongslope topographically induced circulation. Evidence of overflow is found at all locations on the Atlantic flank of the IFR. However, the meridionally oriented portion at 13°W has anomalously warm bottom water and divides FBC and eastern IFR overflow from overflow found in the Western Valley. Individual Seaglider sections identify IFR overflow in a narrow current (8–14 km wide) along the Iceland shelf with a mean transport of 0.43 Sv (1 Sv ≡ 106 m3 s−1) with significant variability from days to weeks. A lower-bound estimate of 0.8 Sv of total IFR overflow is presented. These results provide constraints on regional models that inform the representation of this crucial, yet underresolved, region in large-scale ocean and climate models.

Corresponding author address: Nick Beaird, School of Oceanography, University of Washington, 1492 Boat St., Seattle, WA 98195. E-mail: nlbeaird@uw.edu

1. Introduction

The Greenland–Scotland Ridge (GSR) forms a continuous barrier over which both cold and warm water exchange between the North Atlantic Ocean and the Nordic seas is confined to the upper ocean (≤850 m). Poleward of the ridge, the Nordic seas are filled with cold, recently ventilated waters. To the south, the North Atlantic is warmer and saltier at the same depths. The GSR is split by Iceland into eastern and western regions, with overflow of dense Nordic seas–origin waters occurring on both the western side, though the Denmark Strait, and the eastern side, across the Iceland–Faroe Ridge (IFR) and the Faroe Bank Channel (FBC) (Fig. 1, Table 1). A small volume of dense water also overflows the Wyville Thomson Ridge southeast of the Faroe Islands. The overflows combine with Labrador Sea Water to form North Atlantic Deep Water (NADW), a major constituent of the global abyssal ocean (Johnson 2008). Above the IFR, the warm and cold branches of the Atlantic meridional overturning circulation (AMOC) are squeezed into close contact (Fig. 2). Because the ambient stratification through which a dense plume descends controls many of the plume’s properties, including its eventual detrainment depth (Price and Baringer 1994), the arrangement of cold and warm waters of the AMOC on the IFR has implications for the abyssal circulation of a large part of the Atlantic Ocean. The behavior of these dense overflows is critical to the large-scale ocean circulation.

Fig. 1.
Fig. 1.

(top) Bathymetry and location of features in the area of study. Blue lines indicate the locations of synthetic sections discussed later in the text. Numbered colored squares mark the locations of the profiles in Fig. 3. Thick contours are plotted at 500-m intervals beginning with the 500-m isobath and continuing deeper. (bottom) A profile of ridge crest depth along the dashed line in the top panel with Western Valley (WV), IFR, FBC, and Faroe Bank (FB) labeled.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

Table 1.

Geographic acronyms frequently used in text.

Table 1.
Fig. 2.
Fig. 2.

(top) Salinity contours on a section normal to the IFR occupied by an SG on 3–23 Dec 2007. Dive and climb positions are shown as black vertical tick marks along the upper x axis. Location of the track is shown in the inset map. (bottom) Absolute geostrophic velocity [cross-section component, color contours (m s−1)] and potential density [black contours (kg m−3)]. Positive velocities are toward the northwest.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

The waters around the Iceland–Faroe Ridge have been the focus of intense study dating back to the early twentieth century (Hansen and Østerhus 2000). This paper reports on the mean circulation of eastern overflows that occur across the Iceland–Faroe Ridge and through the Faroe Bank Channel, based on three years of intensive hydrographic surveys by Seaglider (SG) autonomous underwater vehicles. Seagliders collected more than 17 400 hydrographic profiles between November 2006 and 2009 in three-month missions begun quarterly near the Faroe Islands. These surveys greatly increased the quantity of data available on the IFR, especially downstream of the FBC. The spatial resolution of the surveys (from 3 to 6 km) is a considerable improvement over past shipboard studies. We are able to establish our methods by comparing our observations to well-known values (e.g., FBC overflow transport), while the extent and resolution of the Seaglider data allow us to improve understanding of the distribution and transport of the overflows on and across the IFR, establish variability of certain overflow features, and identify regions of dynamical significance to the overflows.

The eastern overflows are principally composed of three source water masses (Fig. 3, Table 2): Norwegian Sea Deep Water (NSDW), Norwegian Sea Arctic Intermediate Water (NSAIW), and Modified East Icelandic Water (MEIW) (Hansen and Østerhus 2000; Fogelqvist et al. 2003). Faroe Bank Channel overflow derives primarily from NSAIW and NSDW sources with a small quantity of MEIW. The IFR overflow contains nearly equal amounts of the three constituents (Hansen and Østerhus 2000). During the overflow process, mixing and entrainment transform the source waters into a product known as Iceland–Scotland Overflow Water (ISOW), a mixture of MEIW, NSDW, NSAIW, and entrained Modified North Atlantic Water (MNAW). The overflow transport roughly doubles because of entrainment and the ISOW attains temperature Θ properties between 2.7° and 2.9°C and salinity S of approximately 34.92 with a potential density σθ ≥ 27.8 kg m−3. Following Hansen and Østerhus (2000), among others, overflow waters will be defined as those with potential density σθ ≥ 27.8 kg m−3, a criterion often used to define ISOW farther downstream.

Fig. 3.
Fig. 3.

Potential temperature vs salinity for six profiles at the location of the numbered colored squares in Fig. 1. Water mass definitions for NAW, MNAW, MEIW, NSAIW, NSDW, and ISOW are shown as gray boxes covering the property ranges of each type. Potential density is contoured in gray.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

Table 2.

Regional water mass property ranges (Fogelqvist et al. 2003; Hansen and Østerhus 2000).

Table 2.

The IFR is approximately 400 km wide with a crest depth varying around 450 m (Fig. 1). In contrast, the FBC is 840-m deep and narrows to 10 km at its primary sill. The eastern overflows are partitioned unequally between the two passages. The canonical estimates of overflow flux are 1.9 ± 0.3 Sv (1 Sv ≡ 106 m3 s−1) and 1 Sv for the FBC and IFR, respectively (Hansen and Østerhus 2007; Hermann 1967). The FBC overflow exhibits a seasonal variation of about 10% of the mean, with a maximum in late summer (Hansen and Østerhus 2007). No estimate of variability is available for the IFR overflow.

The FBC and IFR overflows have different dynamical constraints. The FBC is narrow and deep with a horizontal constriction width on the order of the local first-mode Rossby radius of deformation and likely exhibits hydraulic control (Girton et al. 2006; Pratt et al. 2007; Enmar et al. 2009). The IFR, on the other hand, is very broad compared to the Rossby radius, with the main flow oriented along the ridge axis, controlled by the orientation of the Iceland–Faroe Front (Fig. 2). Cross-ridge flow occurs by eddy flux or in small topographic depressions (Steele 1959; Hansen and Meincke 1979; Hermann 1967; Allen et al. 1994). One of these depressions, referred to as the Western Valley, is located where the IFR joins the Iceland shelf and was repeatedly occupied in this survey, revealing large overflow transport variability.

After passing to the Atlantic side of the GSR, each overflow turns to its right and begins to flow along isobaths under the influence of the pressure gradient and Coriolis forces. The overflow appears as a bottom-intensified current on the Atlantic flank of the IFR (AIFR) with capping isopycnals lying approximately parallel to the bottom slope (Fig. 2) in geostrophic balance with the alongslope current. The vertical density structure over the AIFR includes a thick [O(100 m)] transitional pycnocline between weakly stratified Atlantic and overflow layers (see, e.g., Fig. 10, described in greater detail below). The overflow layer is typically ~100-m thick and often includes a 10–50-m-thick well-mixed bottom boundary layer, in which bottom stress on the overflow reduces the alongslope velocity creating an imbalance between the Coriolis and pressure gradient forces. In the resulting Ekman balance, the alongslope vertical shear stress permits the geostrophically adjusted plume to have a downslope velocity component in the boundary layer (Gill 1982). Eventually, the ISOW will descend from sill depths of 500–800 to 2000–2500 m as it circulates cyclonically in the Iceland Basin (Harvey and Theodorou 1986; Saunders 1996; Fogelqvist et al. 2003), crosses the Reykjanes Ridge, and joins the Denmark Strait overflow in the Irminger Basin. A small portion may recirculate in the Iceland Basin (Harvey and Theodorou 1986). The rate of descent is intimately related to the levels of turbulent bottom stress in the plume bottom boundary layer. The overflow entrains ambient waters all along its path, but by far the greatest changes in properties occur in the first few hundred kilometers from the sill where turbulence-driven diapycnal mixing is highest and plume/ambient property differences are largest (Swift 1984; Price and Baringer 1994).

This paper assesses the three-year-mean flows as well as the variability of overflow plumes in terms of their instantaneous thickness and temperature. Whether the plumes remain continuous, or break into a series of lenses or eddies as they flow away from the sill, remains an important dynamical question. Observations show a strong oscillation of the FBC overflow with a period of approximately 88 h (Geyer et al. 2006; Darelius et al. 2011) indicating spatial and temporal variability in the plume. High-resolution regional models suggest that the FBC overflow plume breaks into eddies (Riemenschneider and Legg 2007; Seim et al. 2010). Observations presented here (see section 5c) and by others (Prater and Rossby 2005) suggest that farther downstream on the IFR the plume does not appear as a series of isolated propagating eddies.

2. Data

Observations considered here were obtained using Seaglider long-range autonomous underwater vehicles. Seagliders are small buoyancy-driven vehicles that profile as deep as 1000 m in a sawtooth pattern, ascending and descending with a typical vertical–horizontal glide ratio of 1:3 (Eriksen et al. 2001). The Seagliders carried Sea-Bird Electronics conductivity (SBE 4) and temperature (SBE 3) sensors, a WET Labs BB2FVMG optical puck with two wavelengths of backscatter as well as fluorescence, and an SBE 43 oxygen sensor. The accuracy of the salinity measurement is 0.01. Corrections are applied to the temperature and conductivity measurements to compensate for thermal-inertia and flushing-speed issues arising from the unpumped conductivity and temperature sensor (C C. Eriksen 2013, unpublished manuscript). Between surfacing, the vehicle navigates by dead reckoning based on a flight model that makes use of an onboard compass and three-axis accelerometer. An estimate of depth-averaged horizontal current (DAC) may be obtained from the vector difference between the dead-reckoned displacement over a dive/climb cycle and the geographic displacement based on GPS fixes at the start and end of each cycle. Thus, each Seaglider sawtooth (or dive cycle) results in two slant-vertical profiles of water column properties and one estimate of DAC. During the Faroes mission, conductivity and temperature were sampled at 20-s intervals. With typical vertical vehicle speeds of 6–10 cm s−1, the vertical resolution was approximately 1.2–2 m. The horizontal separation of these slanted profiles varied with depth, but adjacent Seaglider surfacing locations are separated by 3–6 km. An acoustic altimeter mounted forward on the Seaglider was used to detect and avoid the seafloor. To fully sample the overflow plume, gliders in this survey were programmed to begin their transition from dive to climb 10 m above the acoustically ranged bottom depth. In practice, they turned within a few meters of the seafloor and occasionally collided with it.

Between November 2006 and 2009, 23 successful Seaglider deployments were made resulting in ~17 400 profiles of temperature, salinity, oxygen, red and blue backscatter, fluorescence, and DAC in the Iceland–Faroes region. Of these, 16 deployments were on the AIFR and will be discussed in this paper. The remainder of the deployments was made either north of the Faroes or in the Faroe–Shetland Channel. The average deployment length was 76 days (set by ship’s schedule), covering about 2000 km. Seaglider deployments and recoveries were made quarterly from the research vessel (R.V.) Magnus Heinason during the regular hydrographic surveys conducted by the Faroe Marine Research Institute (FMRI). On average, two Seagliders were deployed on each cruise.

A plot of bottom temperature (Fig. 4) shows the distribution of glider dives. Seagliders were typically deployed near the FBC where they remained for a few days as the shore-based pilot trimmed flight parameters to adjust glider pitch, roll, and buoyancy control. Gliders were then flown along a target isobath of the AIFR, from the FBC in the southeast to the Iceland shelf break in the northwest. Most missions targeted isobaths between 600 and 900 m on the AIFR, and occasionally ridge-normal sections were occupied, as can be seen in Figs. 2 and 4. Strong depth-averaged currents on the IFR made precise repeat sections impossible.

Fig. 4.
Fig. 4.

Bottom temperatures (°C) on the Iceland–Faroe Ridge from all SG dives between November 2006 and 2009. Shaded bathymetric contours are plotted every 250 m, with thick contours every 500 m.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

3. Methods

To describe the transport and structure of the eastern overflows in the multiyear mean, 13 composite sections were created running the length of the IFR from the mouth of the FBC to the Iceland shelf break (Fig. 1). Each section spans the 600–900-m isobath and is oriented approximately parallel to the local bathymetric gradient so that cross-sectional geostrophic velocities would be parallel to the plume velocities. On the AIFR, a substantial part of the FBC outflow descends below 900 m into the Iceland Basin. Therefore, the composite sections do not capture the entire width of the plume beyond the FBC terminus. Some overflow is lost between consecutive sections owing to downslope Ekman drainage and some overflow is gained from the IFR crest.

Sections of potential density, temperature, and salinity were calculated as follows. Each section has a horizontal grid spacing of 500 m and vertical spacing of 10 m. A single gridpoint average is found by taking all Seaglider data at the same depth within a radius L. These data were then subsampled by including only observations below which the bottom depth is within a vertical height H of the bottom depth at the grid point. Subsampling by the bottom depth criterion reduces the cross-slope length scale of averaging. When averaging on depth surfaces, this anisotropy is desirable because dense overflow isopycnals are draped over the AIFR, lying nearly parallel to the sloping bottom (Fig. 2). All observations that meet the above criteria are averaged to obtain an estimate at the grid point. For sections on the AIFR, L = 15 km and H = 40 m. With typical IFR slopes of 1/100, H = 40 m implies a cross-isobath averaging length of 4 km. In the Faroe Bank Channel region (sections I–III) and at the Iceland shelf break (section XIII), the averaging length scales, L = 5 km and H = 100 m, are chosen to better suit the steeper topography and narrower current features observed in these regions.

Profiles of relative geostrophic velocity normal to each composite section were calculated from the average density. Reference velocities were created by averaging the Seaglider estimates of DAC using the H and L criteria described above. The depth average of the relative geostrophic velocities at a grid point was adjusted to match the cross-sectional component of the average DAC at that point, resulting in a profile of absolute geostrophic velocity. To account for reductions to alongslope transport by friction near the bottom, an Ekman layer was imposed on the velocity profiles in a bottom boundary layer. It is assumed that the near-bottom geostrophic flow is predominantly perpendicular to the sections. An estimate of Ekman layer thickness was made by averaging the bottom mixed layer (BML) depth from each of the density profiles associated with a grid point (typically 10–50 m thick). We subtract an average Ekman layer velocity from the profile of geostrophic velocity in the boundary layer, given by
e1
where Cd = 2.5 × 10−3 is a drag coefficient, Utbl is the geostrophic speed at the top of the boundary layer, δ is the Ekman layer thickness, and f is the Coriolis parameter. This is simply the textbook Ekman transport with a quadratic drag parameterization and divided by the layer thickness to get an average velocity.

Uncertainty of the transport through the composite sections was assessed by a Monte Carlo approach. We consider the uncertainty of the components of the geostrophic velocity: the cross-sectional component of DAC and the averaged density field (connected via thermal wind to velocity). We calculate the transports of an ensemble of sections with perturbed density and reference velocities, in which perturbations are based on the uncertainties of the average quantities. The procedure is as follows: a single section in broken into a group of smaller segments each 4 km wide. The number of glider dives that falls within the end points of each segment is used as the number of degrees of freedom (DOF) for standard error estimates. Two sets of random normal variables (100 values each) are produced for each 4-km segment, one to perturb the DAC reference velocity and one to perturb the densities. At each grid column in the segment (there are eight columns per segment), an ensemble of perturbed reference velocities is created by taking the mean cross-track component of DAC, and adding to it the random normal distribution multiplied by the standard deviation of the DAC divided by the square root of the DOF. Thus, at each grid column, there is a distribution of reference velocities with the statistics of the mean value and its standard error. The choice to use a 4-km-wide segment with the same random number distribution was made to mimic correlation between adjacent points and to be consistent with the originally used averaging scheme. The second random normal distribution is used in the same way (mean plus random distribution multiplied by the standard error) at every grid point (x, z) to create an ensemble of perturbed densities. The result is 100 sections of randomly perturbed density and reference velocity. We calculate the absolute geostrophic velocities and transport for each one, and report the standard deviation of the results as uncertainty bars on the mean transport estimates.

Additionally, the DAC observations, whose distribution is that of the points in Fig. 4, are statistically interpolated onto a regular grid to produce the near-bottom circulation map (shown in Fig. 12, described in greater detail below). Statistical interpolation, described by (Bretherton et al. 1976), relies heavily on the spatial covariance between the data and itself and between the data and the grid points to produce an estimate at a grid point given by a weighted linear combination of all observations.

4. Faroe Bank Channel

The Faroe Bank Channel is the deepest gap in the GSR, and roughly two-thirds of the flux of eastern overflow waters passes through this narrow channel. The following section compares Seaglider observations of the long-term structure of the overflow with some of the numerous previous FBC studies. Transport estimates made assuming geostrophic balance compare well with the estimates of Hansen and Østerhus (2007), who utilized long-term moored velocity measurements upstream at the FBC sill. The increased spatial resolution of the Seaglider data defines in greater detail the structure of the FBC overflow downstream of the sill, indicating dynamically significant regions of plume thinning and bifurcation.

a. FBC composite section

Synthetic section I is located approximately 20 km downstream of the sill (Fig. 5a), near section E of Mauritzen et al. (2005), section B of Fer et al. (2010), and section P of Duncan et al. (2003). Figure 5 shows plots of DAC, potential temperature, and absolute geostrophic velocity along the section. Because of the averaging method described above, transport and overflow characteristics should be interpreted as reflecting the mean conditions 5 km up- and downstream of the section in Fig. 5a.

Fig. 5.
Fig. 5.

(a) Location of section I in FBC and averaged DAC vectors (red) used to reference geostrophic velocity calculations. Depth contours at 100-m intervals with 500-m contour shown by thick line. Depth (m) by cross-channel distance (km) sections of (b) potential temperature (°C) and (c) absolute geostrophic velocity (m s−1; positive to the northwest). The mean depth of the 3°C isotherm observed in individual dives is plotted in black with dashed lines indicating plus or minus one std dev.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

The temperature section (Fig. 5b) displays the classic shape of the FBC overflow with deeper isotherms on the Faroe Bank (FB) side of the channel sloping upward toward the Faroe Plateau (FP) side. In addition to the isotherm tilt, past synoptic sections have revealed interface stratification that is higher on the Faroe Bank side than on the Faroe Plateau side of the channel (Hansen and Østerhus 2000). This isopycnal pinching, attributed by Johnson and Sanford (1992) to a secondary frictionally driven helical flow, is partially obscured by averaging in the synthetic section. However, the mean buoyancy frequency at the 3°C isotherm averaged from individual profiles decreases from 4.3 cycles per hour (cph) on the Faroe Bank side to 2.7 cph at the Faroe Plateau side (not shown) of the channel. The cross-channel gradient in the interfacial stratification is a result of the secondary circulation in which bottom water is advected to the left of the main overflow (looking downstream) in a frictional Ekman layer and returned by an interior transverse flow balanced by downstream isotherm tilt (see Fig. 4 in Seim and Fer 2011). At the Faroe Plateau edge of the plume, the tendency for the Ekman transport to pull less-dense interfacial waters into the bottom Ekman layer may lead to overturning and increased mixing (Seim and Fer 2011).

Depth-averaged flow in the FBC is oriented along the channel with speeds up to ~30 cm s−1 (Fig. 5a). The high speeds of the baroclinic overflow dominate the DAC. However, Fig. 5c also shows some barotropic flow. Cyclonic shear across the channel is apparent in the layer above the overflow, with more upstream flow on the Faroe Bank side of the channel, as has been reported previously (Hansen and Østerhus 2007). The total-mean geostrophic transport of waters denser than 1027.8 kg m−3 through section I is 1.8 ± 0.2 Sv, where the uncertainty comes from the standard deviation of the Monte Carlo results described in the methods. The estimate is in agreement with the long-term average of 1.9 ± 0.3 Sv measured at the sill by moored ADCPs (Hansen and Østerhus 2007), though it might have been expected that entrainment would increase the transport at section I relative to the primary sill transport (20 km upstream). Averaging is done on depth surfaces and, despite the attempt to compensate for the shorter cross-slope decorrelation length by using the H criterion, horizontal averaging works to flatten isopycnal tilt. Reducing the averaging length scale increases the cross-channel isopycnal tilt, but drastically reduces the data available for the composite section. As a consequence, the overflow transport estimate reported here is a lower bound for the 3-yr mean.

Section I transport may also be compared with the synoptic section E of Mauritzen et al. (2005) and B of Fer et al. (2010), each of whom uses σθ ≥ 27.65 kg m−3 to define the overflow. Mauritzen et al. (2005) estimate the instantaneous overflow to be 2.0 Sv, which is slightly lower than our long-term estimate of 2.35 Sv (recalculated for waters with σθ ≥ 27.65 kg m−3). Fer et al. (2010) also report a transport of approximately 2 Sv at their section B. The transport weighted–mean ISOW (σθ ≥ 27.8 kg m−3) properties at section I are Θ = 1.11°C, S = 34.96, and σθ = 27.98 kg m−3. Section I shows the effects of entrainment of MNAW diluting the mean plume values at the sill reported by Hansen and Østerhus (2007): Θ = 0.25°C, S = 34.93, and σθ = 28.01 kg m−3. Seaglider vertical velocity observations that were used to infer dissipation of turbulent kinetic energy in the FBC (Beaird et al. 2012), revealed a hot spot of mixing between the primary sill and composite section I responsible for the observed dilution of mean plume properties.

b. FBC exit

The FBC overflow plume encounters a secondary sill at the northwestern end of the FBC, beyond which the overflow spills onto the Atlantic side of the IFR. The plume undergoes significant changes in the region between the secondary sill and a topographic prominence approximately 65 km downstream (bump, Fig. 6). In this region, the plume thins, bifurcates, and encounters elevated mixing.

Fig. 6.
Fig. 6.

(top) Thickness (m) of the ISOW layer near the FBC exit based on a fine-grid objective map of all SG data. Plume thickness is the vertical distance between the mapped σθ = 27.8 kg m−3 isopycnal surface and the seafloor. Bathymetric contours at 50-m intervals with 1000- and 500-m isobaths shown by thick lines. The deepest point is indicated by a black-and-white dashed line. The portion of the 1000-m isobath where the downslope plume edge is deeper than the reach of the SG is highlighted in pink, indicating the area through which the deeper branch of the plume flows. The primary and terminal (second) sills of the FBC are labeled, along with the topographic bump, which marks the end of the plume bifurcation region. (bottom) Near-bottom temperatures (°C) from SG dives in the FBC outflow region. The bathymetry is the same as the top panel.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

An objective map of all Seaglider plume thickness observations (Fig. 6, top) illuminates several changes that occur as the overflow exits the FBC. Within the channel, the overflow plume is over 200 m thick, with the thickest part of the plume on the Faroe Plateau side of the channel. At the secondary sill (about 9°W), the plume thins to approximately 115 m and spreads horizontally. This thinning corresponds to the point at which the plume flows onto the face of the IFR, evolving from an enclosed channel flow to a gravity current on a slope. Thinning and widening of the plume qualitatively suggests hydraulic control at the secondary sill. Plume widening is also seen in the models of Riemenschneider and Legg (2007) and Seim et al. (2010), and is attributed in the former to a transverse hydraulic jump (Pratt et al. 2007). Using three techniques for calculating Froude numbers, Girton et al. (2006) find subcritical flow in much of the plume, observing supercritical values only in sections at, and just beyond, the secondary sill.

Between the secondary FBC sill and the bump, the overflow plume bifurcates into a shallow and deep branch. A portion of the overflow is funneled around the bump to the north, and a larger portion flows to the south into deeper water (across the pink contour in Fig. 6). The shallow pathway is seen as a relatively thick branch (≥100 m) extending northwest from the bump in Fig. 6. The continuation of the deeper branch is not shown in Fig. 6 because it is below the maximum dive depth of Seagliders. Thus, downstream of the intersection of the plume edge and the 1000-m isobath (pink contour in Fig. 6), the thickness map does not represent the full extent of the overflow [for observations of the deeper branch see section S in Fig. 7 in Duncan et al. (2003), and Fig. 3:108 in Hermann (1967)]. The majority of the coldest overflow from the FBC follows the deeper branch. This is clear from the bottom temperature distribution in the bottom panel of Fig. 6, where the coldest observations downstream of the FBC tend to follow a trajectory to the south of the topographic bump. In section 5a, it will be shown that roughly 0.6 Sv (about one-third of the transport at composite section I) flows in the shallow branch.

The region of thinning, widening, and bifurcation corresponds well to observations of enhanced mixing identified in discrete sections by Mauritzen et al. [(2005), 100 km downstream of the primary sill] and Fer et al. [(2010), ~80 km]. Analysis of Seaglider vertical velocity data (Beaird et al. 2012) indicates that turbulent dissipation in the overflow plume is greatly enhanced in the region between the secondary sill and the topographic bump. This region of elevated dissipation is larger than, and distinct from, another located just downstream of the primary sill. Observations of Girton et al. (2006) and Fer et al. (2010) as well as model results of Riemenschneider and Legg (2007) find critical Froude numbers in this region and suggest mixing produced by a wave regime as found in the laboratory by Cenedese et al. (2004). Figure 6 shows that bottom temperatures rise from around −0.5°C near the FBC primary sill to over 0°C just beyond the secondary sill (purple–blue transition). Much of the change occurs at and immediately beyond the second sill, evidence of increased entrainment of warm Atlantic water.

The abrupt changes that occur downstream of the terminus of the FBC suggest that, perhaps unsurprisingly, the secondary sill marks the most important transition region for the FBC overflow, where the majority of entrainment, water mass modification, and dynamic adjustment occur.

5. Iceland–Faroe Ridge

Beyond the Faroe Bank Channel the overflow plume spills out onto the Iceland–Faroe Ridge, which forms the northern boundary of the Iceland Basin. As the FBC overflow travels downstream, it gradually descends deeper into the Iceland Basin as a consequence of frictional downslope transport. Observations of the pathways of descent are scarce, but a plausible picture emerges in the model results of Xu et al. [(2010), see their Fig. 7]. By the time it reaches the slope south of Iceland, the overflow is found between 1300- and 2300-m depth (Saunders 1990; Fogelqvist et al. 2003). Along the crest of the IFR, the FBC plume is joined by the weaker and more-intermittent IFR overflow. It is difficult to distinguish between the FBC and the IFR overflows because each shares similar source waters, having been mixed with the same overlying MNAW.

a. IFR composite sections

Figure 7a depicts the layers of ISOW density for each of the 13 composite sections shown in Fig. 1. Each section is offset by 300-m depth. The figure indicates that, in the mean, isopycnals associated with dense overflow waters lie approximately parallel to the topographic slope. The orientation of the isopycnals with respect to the ridge slope produces shear leading to bottom-intensified alongslope geostrophic flow in the overflow layer.

Fig. 7.
Fig. 7.

(a) Mean overflow potential density contours and (b) observations of the depth of the 4.5°C isotherm for each of the 13 composite sections shown in Fig. 1. Sections are offset by 300-m depth for clarity, with vertical tick marks placed every 100 m. Labeled depths refer to section I. Section roman numerals are centered at 600 m in the reference frame of each section. Dark lines indicate bathymetry along each section. Width is measured with respect to the location of the 700-m isobath. In (b), black dots show observations of the depth of the 4.5°C isotherm while the thin gray line shows the mean depth.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

Figure 7b shows the depth of the 4.5°C isotherm (σθ ≈ 27.75 kg m−3) from individual dives. Variability of plume thickness about the mean is approximately constant across the slope. Substantial variability exists in the thickness of the overflow layer. However, the general arrangement of isotherms parallel to bottom slope is a robust feature discussed later in section 5c.

The first three sections in Fig. 7a indicate a rapid downstream decrease in the thickness of the ISOW layer. This decrease occurs as the overflow leaves the FBC, mainly between sections II and III, which are located on either side of the secondary sill of the FBC. Large mixing and entrainment erode the densest waters. The first three sections are located upstream of the bifurcation. Thus, they cover the entirety of the overflow and any changes in mean properties must be due to entrainment of overlying MNAW (individual bottom temperature observations find the plume edges to be inside the bounds of these sections at all times; Fig. 6). The bifurcation occurs between sections III and IV, where only a portion of the overflow remains above the 1000-m isobath. A significant loss of the densest waters between sections III and IV suggests that the densest part of the overflow follows the deeper branch (as in Fig. 6). Downstream of section IV, thickness and mean density of the ISOW layer decline until section XI, after which the ISOW layer grows as a result of the addition of IFR overflow water. The overflow layer is thinnest at sections X and XI. This part of the AIFR, just beyond the bend near 13°W where the ridge becomes meridionally oriented, contains only a thin layer of overflow water.

Transport of overflow waters through all the composite sections is plotted in Fig. 8. Included in the figure are transport estimates from various other studies, with error bars where included by the authors. The bounds on the composite sections are the result of the approach described in the methods section. Assuming that the overflow is entraining, the transport should increase downstream between sections I, II, and III, where the entire plume width is covered. Figure 8, however, unexpectedly shows a decrease in transport at sections II and III. This decrease occurs in the region at, and downstream of, the secondary sill where dissipation is enhanced and hydraulic control takes place (Girton et al. 2006; Pratt et al. 2007; Fer et al. 2010; Beaird et al. 2012). The decreased transport through sections II and III suggests that the turbulent plume is not in geostrophic balance in that region, and thus the assumptions made to obtain these transport estimates are not valid. The decrease occurs even when lighter ISOW density definitions are used. Thus, the decrease is not due to increased entrainment and transport in slightly lighter waters.

Fig. 8.
Fig. 8.

Cross-sectional transport through composite sections I–XIII (SG composite) vs distance from FBC sill. Std dev from the Monte Carlo approach are shown by the pink patch. The average transport through individual SG sections in the WV is shown as well. Included are transport estimates from various authors: blue symbols indicate estimates using the σθ ≥ 27.65 kg m−3 criterion, whereas red symbols use σθ ≥ 27.8 kg m−3. Black markers indicate other plume definitions used by various authors [model tracers in Riemenschneider and Legg (2007), depth of no motion in Perkins et al. (1998), and waters colder than 3°C in Darelius et al. (2011)].

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

Between composite sections IV and IX, transport remains relatively constant at about 0.6 Sv; only about 33% of the total FBC overflow transport at composite section I remains above 900-m depth after the bifurcation between sections III and IV. This constitutes the shallow overflow branch mentioned previously. The remaining 67% of the FBC plume flows into the deep branch. The sections show the geostrophically balanced transports with a reduction in the boundary layer owing to the modeled Ekman layer. The frictional ageostrophic flow in the bottom boundary layer should be directed downslope into the Iceland Basin. This downslope flow can be estimated by integrating the Ekman balance equation over the boundary layer thickness and assuming the stress at the top of the layer is zero, leading to a downslope transport per unit width of
e2
where the bottom stress has been parameterized by a drag coefficient (i.e., Cd) and the square of the overlying alongslope speed (i.e., Utbl) such that . A rough calculation assuming Cd = 2.5 × 10−3 and Utbl = 21 cm s−1 (the AIFR mean, see section 5c) suggests that the downslope Ekman drainage over the 146 km between sections IV and VIII is about 0.12 Sv. Thus, assuming alongslope flow, in the absence of additional overflow sources the transport between IV and VIII would decrease by 0.12 Sv. However, Fig. 8 shows a possible increase of 0.2 Sv, implying about a 0.3 Sv addition of IFR overflow. Between sections VIII and XI, the transport diminishes from ~0.75 to ~0.15 Sv indicating that most of the shallow branch of the FBC overflow and some accumulated overflow from the eastern IFR descend deeper into the Iceland Basin in this region. Between sections XI and XIII, the dense transport increases by 0.5 Sv. Because the overflow plumes must descend in the mean, any convergence of alongslope flow like this implies additional overflow from the IFR crest. We conclude from the composite sections that at least 0.5 Sv of overflow crosses the western portion of the IFR (sections XI–XIII) in the mean, and the above calculation suggests about 0.3 Sv of overflow across the eastern IFR (sections IV–VIII), bringing the total IFR overflow estimate to 0.8 Sv. This is considered an underestimate of the total IFR overflow as we are calculating the IFR overflow as a residual from convergences in alongslope flow and a simple downslope Ekman layer. Perhaps other ageostrophic contributions to the IFR overflow are missed, but the 0.8-Sv estimate compares moderately well with the (highly uncertain) canonical estimate of 1 Sv for IFR overflow (Hermann 1967).

b. Distribution of overflow on the IFR

Figure 9 shows the mean bottom potential temperature (left) and minimum potential temperature anomaly (right) from the mean in 7.5 km × 7.5 km boxes on the AIFR. In the mean, the FBC overflow stands out clearly, with a bottom layer colder than 1°C extending to the topographic bump near 10°W. The coldest overflow, with temperatures ≤0°C, is, in the mean, diluted downstream of the primary FBC sill. A layer colder than 3°C stretches downstream to about 12°W. The meridional arm of the IFR (at about 13°W) is warmer, with the near-bottom temperatures averaging between 3° and 5°C. Maps of overflow-layer thickness from three surveys of the OVERFLOW ′60 experiment likewise show the meridional portion of the IFR with a thin covering of cold water (Hermann 1967, Fig. 3:108). Previous studies have identified the northwesternmost depression in the IFR, the Western Valley, as a site of frequent overflow (Perkins et al. 1998). The appearance of a thin strip of (relative to regional means; ≤3°C) low-mean temperatures along the 500-m isobath of the Iceland shelf break provides evidence for this pathway of IFR overflow. This ribbon of low-temperature overflow is isolated from similar temperatures upstream, to the southeast, by the high-mean-temperature region of the meridional arm of the AIFR. Thus, the source of the shelfbreak overflow must be local, implicating the WV region.

Fig. 9.
Fig. 9.

(left) Mean bottom potential temperature (°C) in 7.5 km × 7.5 km boxes. (right) Min bottom potential temperature anomaly from the mean in the same boxes (°C). Numbered and outlined boxes in the left panel correspond to the profiles in Fig. 10.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

The minimum temperature anomaly field in the right panel of Fig. 9 adds information about the variability of the overflow circulation on the AIFR. The minimum temperature anomaly in the FBC is small compared with the rest of the AIFR. The large minima along the southwestern edge of the FBC are due to the coarse spatial bins overlapping the relatively sharp plume edge in that region. On the AIFR outside the FBC, the magnitude of the minimum temperature anomalies increases, indicating variability in the amount of mixing and entrainment that has taken place in the plume at a particular location. The Western Valley exhibits the coldest minimum anomalies, reflecting strong variability of the WV overflow. Figure 9 agrees with the results of mean sections reported by Meincke (1972), indicating warmer near-bottom temperatures on the meridional portion of the IFR and enhanced variability on the eastern half of the IFR and at the Iceland shelf break.

Potential temperature–depth profiles and temperature–salinity (Θ–S) plots from the four boxes outlined in black in Fig. 9 are shown in Fig. 10. These profiles are included to show instantaneous profiles and point out variable and persistent features of the overflow. In the background of each panel are profiles from the upper 800 m of both the Nordic seas (blue), representing the source of the overflow, and from the Iceland Basin (red), showing ambient conditions around the plume on the AIFR. The boxes are chosen arbitrarily, but each is along the 750-m isobath and represents a different region of the IFR: the FBC (box 1), a mid-IFR location (box 2), the meridional arm region (box 3), and the Western Valley (box 4). Profiles in each box all have the characteristic deep pycnocline of the overflow plume implying that if the overflow plume propagates as a series of isolated eddies or lenses of cold water [as in Nof et al. (2002)], these lenses are features superposed on a thin, continuous, sheet of overflow.

Fig. 10.
Fig. 10.

(top) All potential temperature profiles and (bottom) all Θ–S curves contained in the four boxes outlined in black in Fig. 9. From the FBC box on the right (box 1) to the left, the boxes move downstream. The thick contour on Θ–S plots is the 27.8 kg m−3 isopycnal with 0.1 kg m−3 intervals. Blue profiles show upstream conditions in the Nordic seas and red profiles show ambient conditions in the Iceland Basin.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

The intensity of the stratification varies, as does the temperature and mixed layer depth of the plume layer. Taking the BML to be the height above bottom (HAB) where the temperature differs from the bottommost value by 0.1°C, the stratification in the interfacial layer (0.05°–0.1°C m−1) ensures that BML thickness, even with this coarse definition, will be overestimated at most by a few meters, a small percentage of BML thickness anywhere in the FBC or on the IFR. The profiles of box 1 show thick BMLs (~150 m), low BML temperatures (from −0.4° to −0.1°C), and a thick interfacial pycnocline (~200 m). Box 1 has little temperature variability with respect to downstream locations. However, BML thickness varies from 125 to 180 m. Boxes 2–4 exhibit enhanced plume temperature variability. Plume temperature is at a maximum in box 3, while box 4 clearly contains recent cold overflow from the IFR. The appearance of lower salinities at intermediate density in boxes 2–4 suggests the presence of less-modified low-salinity MEIW and overflow from the IFR crest.

Several temperature profiles in boxes 2–4 appear colder by nearly 2°C in the upper 400 m than the rest of the profiles. The colder profiles have been taken within cold core eddies that emanate from the Iceland–Faroe Front. The eddies have a cold surface expression, colder near-bottom temperatures, and layers of low salinities. The role these eddies play in the exchange of North Atlantic and Nordic seas water across the IFR will be investigated in a separate paper.

Mean and variability of the near-bottom-layer properties on the AIFR between 500- and 1000-m depth are presented in Fig. 11. The figure shows the mean and standard deviation of the bottom mixed layer temperature (top), and height above bottom of the 4.5°C isotherm (bottom) plotted against distance from the FBC sill. Bottom mixed layer temperatures on the AIFR are around 3°C, rising on the meridional portion of the AIFR (300–375 km), falling again in the WV (375–400 km). The HAB of the 4.5°C isotherm, a good representation of the overflow-layer thickness, on average is between 75 and 100 m on the AIFR. The standard deviation of BML temperature is about 33% of the mean and about 50% for overflow thickness. These remain fairly constant on the AIFR outside the FBC.

Fig. 11.
Fig. 11.

Bottom-layer property means (solid line) and std dev (dashed line) on the IFR between the 500- and 1000-m isobaths in 15-km bins of distance from the primary FBC sill. Individual observations are plotted in light gray. (top) BML temperature. (bottom) HAB of the 4.5°C isotherm. The vertical gray lines around 50 and 80 km indicate the location of the secondary FBC sill and the point at which the FBC overflow first intersects the 1000-m isobath, respectively.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

c. Near-bottom velocities

The near-bottom-mean velocity field over the Atlantic flank of the IFR is shown in Fig. 12. Absolute geostrophic velocities were calculated from a mean density field averaged in 7.5 km × 7.5 km by 10-m tall boxes. Shear was computed from horizontal density gradients and integrated to produce a vertical profile of relative velocities, the depth average of which was then matched to a statistically interpolated field from the Seaglider DAC observations. The absolute geostrophic velocities in Fig. 12 are plotted 75 m above the bottom in an effort to both be within the overflow plume and avoid the frictional boundary layer in which geostrophic balance does not hold. Qualitatively, the velocities confirm the canonical near-bottom circulation scheme on the AIFR, with along-isobath flow from the southeast to northwest becoming southwestward at the Iceland continental shelf (Hansen and Østerhus 2000). This circulation reflects the parallelism of isopycnals and ridge topography seen in Fig. 7. This field can be compared with the tracks of Prater and Rossby (2005), found from bottom-following RAFOS floats drifting approximately 100 m above seafloor. The three floats passed into the deep branch of the bifurcation, remaining with the majority of the FBC overflow generally deeper than 1000 m. Quantitatively, the measured geostrophic velocities support the trajectories of the Prater and Rossby floats (gray markers in Fig. 12). Near-bottom geostrophic velocities average 18–30 cm s−1 with maxima of 60–70 cm s−1 near the Faroe Bank Channel outflow (below 600-m isobath, mean: 21.5 cm s−1, standard deviation: 11 cm s−1). These estimates agree well with the mean speeds of the RAFOS floats, typically 20–30 cm s−1 with peaks of 40–50 cm s−1 (Prater and Rossby 2005). The maximum speeds in the FBC (60–70 cm s−1) are lower than expected and are a consequence of the coarse 7.5 km × 7.5 km grid, which is too large to capture horizontal density gradients in the narrow FBC. Both the geostrophic velocities and the float trajectories suggest the smoothness and continuity of the near-bottom flow. However, this mean flow is embedded in the well-documented strong eddy field of the region (Allen et al. 1994). The root-mean-squared DAC over the entire IFR is 17.15 cm s−1, with higher values on the ridge shallower than the 500-m isobath and in the FBC outflow. The standard deviation of the DAC is 11 cm s−1. The variability in the DAC is smaller than the average near-bottom current speed, which helps to explain the direct trajectories of the Prater and Rossby (2005) RAFOS Floats.

Fig. 12.
Fig. 12.

Near-bottom absolute geostrophic velocities (75 m above bottom) and RAFOS float trajectories (Prater and Rossby 2005). Vector color indicates potential density at 75 m above bottom.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

To assess the importance of mesoscale velocity relative to the mean (alongslope) velocity, we compute the thermal wind shear caused by sloping isopycnals observed between pairs of dives sorted into groups according to whether the glider was heading along or across the IFR slope. The slope of an isopycnal Sisp in the core of the stratified interface between the plume and the ambient Atlantic water is taken to be representative of interface tilt and is assumed to represent the majority of the thermal wind shear. The calculations below use the slope of the 4.5°C isotherm to represent Sisp, an approximation valid in this region where density is dominated by temperature. The Margules relation may be derived from the thermal wind shear to give an expression for the strength of the baroclinic velocity in the plume:
e3
where Δρ and ΔU are the density and velocity difference across the plume interface, respectively. Inspection of cross-ridge sections, like those in Fig. 2, suggests that there is a strong background slope of the isopycnals that is nearly parallel to the topographic slope and perturbed by mesoscale variability. To estimate the relative importance of the background slope and the mesoscale perturbations of the slope to velocities in the overflow plume, pairs of dives on the AIFR are separated into two groups. The cross-ridge (x ridge) group, contains all dive pairs where the magnitude of the bottom slope (as measured by the acoustically ranged bottom depth and dive separation) between the two dives is at least 80% of the local topographic gradient calculated from 1 arc-minute global relief model of Earth’s surface (ETOPO1). On the AIFR, below 500 m and between composite sections IV and XII, 800 dive pairs meet the criteria. The second group, referred to as along ridge (a ridge), consists of dive pairs where the bottom slope is less than 20% of the topographic gradient; 433 pairs met the criteria. The Sisp calculated from the x-ridge group should contain the background isopycnal slope suggested by Fig. 2, as well as some mesoscale signal. The a-ridge group should only contain Sisp associated with mesoscale features, as there is no observable along-ridge-mean isopycnal slope on the AIFR outside of the FBC. Characterizing the isopycnal slope statistics of each group permits a description of the magnitude of eddy-like features relative to the larger-scale topographically induced component of the overflow circulation.

The distribution of isopycnal slopes in Fig. 13b from the a-ridge group is nearly Gaussian and centered at zero, while the distribution of the x-ridge group is shifted to the right, centered at 5 × 10−3. This supports the hypothesis that the along-ridge group samples the eddy-like mesoscale variability of the plume interface slope with zero mean. The x-ridge group contains the same mesoscale variability, along with the background isopycnal slope approximately parallel to the topography seen in Fig. 2. The parallelism becomes apparent in Fig. 13e where the distributions of the ratio of isopycnal to local topographic slope show the x-ridge group centered on one, and the a-ridge group centered on zero. For each group, the density anomalies from an Iceland Basin–mean density profile are shown in Fig. 13a. Calculating the magnitude of the velocity jump for each group from Eq. (3) leads to Fig. 13c. The panel also contains the distribution of velocity jump calculated using the local topographic gradient in place of Sisp in Eq. (3). The mean velocity magnitudes of the a-ridge, x-ridge, and topography groups are 13.6, 35.7, and 34.1 cm s−1, respectively, suggesting that the eddy component of baroclinic velocity is only one-half of that due to the cross-IFR isopycnal tilt. The velocity due to cross-ridge isopycnal tilt is very close to that predicted by the topographic slope, implying that the topographic slope alone provides a good estimate of plume velocity magnitude.

Fig. 13.
Fig. 13.

Distribution of variables related to eddy and mean baroclinic velocities on the AIFR. Thick black lines represent the x-ridge group and medium gray lines represent the a-ridge group in (a)–(e). In (c), the light gray line represents velocities based on topographic slopes. (a) Distribution of vertically averaged plume density anomaly from ambient Iceland Basin density. (b) Isopycnal slopes between dive pairs. (c) Magnitude of the velocity jump across the plume interface due to the isopycnal slope and density anomaly as in Eq. (3). The light gray line shows the velocity jump if Sisp is replaced by the local topographic gradient. (d) Distribution of the along-section BML temperature gradient between sequential dives. (e) Distribution of the ratio of isopycnal to topographic slope. (f) Distribution of the ratio of velocity jumps due to eddy isopycnal slopes to that due to topographic slope for the x-ridge group (thin black line), a-ridge group (gray line), and all dives (thick black line).

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

Assuming that the a-ridge group provides information on eddy-like variability and that SispStopo for the x-slope case, any dive pair may be used to obtain the eddy component of baroclinic velocity by removing the slope due to topography (Seddy = SispSdive depth) and compare its implied velocity to the topographically induced velocity at the dive location:
e4

In Fig. 13f, the ratio of eddy velocity to topographic velocity is shown for the a-ridge group (gray line), x-ridge group (thin black line), and all dives on the AIFR (thick black line). Removal of the bottom slope collapses the x-ridge velocity distribution onto the a-ridge distribution. The distribution of all dives show 80% of the eddy velocity jump magnitudes are smaller than 1, that is, weaker than the topographically induced baroclinic velocities. This analysis suggests that eddy-like mesoscale variability influences baroclinic velocity in the overflow plume, but its strength is only a fraction of the background state, which may be approximated by the topographic slope.

6. Western Valley

At the northwestern end of the IFR, where the ridge meets the Iceland shelf, a relatively deep depression cuts the ridge crest. Referred to as the WV, this region has been identified as a location of at least intermittent overflow across the IFR (Perkins et al. 1994, 1998; Wilkenskjeld and Quadfasel 2005). The observations of bottom temperature upstream of the WV (in the sense of along isobaths in the direction of the FBC) are warmer than those in the WV against the Iceland shelf (Figs. 9 and 14). A decrease in bottom temperature along a given isobath in the WV provides evidence that overflow is entering the region from the direction of the IFR crest. This hypothesis is supported by the properties of the overflow found in the WV, which is relatively cold and fresh, suggesting that it is close to its source and has not had a significant amount of time to mix with the overlying Atlantic water (e.g., Fig. 10, box 4, located in the WV).

Fig. 14.
Fig. 14.

Bottom temperatures from individual dives in the Western Valley showing anomalous cold waters along the Iceland shelf.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

Additional support for western IFR origins of WV overflow comes from analysis of a simple overflow model. The model, developed in Killworth (2001), assumes a quadratic turbulent bottom drag and local equilibrium turbulence solution. Overflows in the model follow a simple trajectory where their rate of descent is constant assuming that turbulence in the bottom boundary layer remains high. To identify the possible origins of the overflow water, each point along composite section XIII was taken as an endpoint of the Killworth model, which was then run backward using a range of descent rates to estimate the starting location of each point. Figure 15 shows the results using a descent rate of 1 m in 600 m of horizontal travel. The trajectories of points along section XIII trace out a watershed of overflow found in the WV, that is, the region of the IFR crest from which overflow would drain into the WV. Different descent rate parameters give qualitatively similar solutions. Namely, overflow found in the Western Valley crossed the IFR through the WV proper or on the northwestern half of the IFR. Steeper descent rates (1/400, 1/200) restrict the WV overflow source farther to the northwest.

Fig. 15.
Fig. 15.

Trajectories of particles arriving at section XIII (Fig. 1) computed using the constant descent rate (1/600) model of Killworth (2001). The model is run backward for 300 km from the blue points along section XIII. Color is an index of the position along section XIII of each arriving trajectory.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

Where the IFR abuts the Icelandic Plateau, the Iceland shelf becomes the dominant topographic feature. Overflow isopycnals lean against the Iceland slope as the flow changes direction toward the southwest, flowing away from the IFR toward the Reykjanes Ridge. Current meter records in the region indicate that long-term near-bottom velocities are steady and directed along the Iceland shelf with mean speeds of 50 cm s−1 (Perkins et al. 1998). It is unclear from the Perkins et al. (1998) study if the steady near-bottom velocities imply a steady overflow transport as contemporaneous temperature records are not published with the current meter data. Multiple Seaglider transects in the region indicate the overflow transport in the WV is more variable than the current meter records of Perkins et al. (1998) suggest.

The Seagliders in this study made 28 transects across the WV between 2006 and 2009. An attempt was made in each transect to approach the Iceland shelf normal to the isobaths so that cross-sectional transport estimates of alongslope flow could be made using geostrophy. Shelfbreak currents and eddies distorted the sections somewhat but rough estimates of alongslope overflow transport can be made. Very often a narrow current of dense water, between 8 and 14 km wide, was found leaning against the Iceland shelf, the density and thickness of which varied considerably. Figure 16 presents four example density sections, two of which have large overflow transport, and two of which exhibit very little overflow. The figure contours all density anomalies greater than 27.65 kg m−3, but again we use the σΘ ≥ 27.8 kg m−3 criterion to define the overflow transport of the Iceland shelfbreak overflow current. Between 26 June and 7 July 2009, a single Seaglider made four sections normal to the Iceland shelf break (not pictured). The Seaglider found a shelfbreak overflow current with transports of 0.44 (27–28 June), 0.68 (28–30 June), 1.17 (30 June–1 July), and 0.07 Sv (6–7 July). Though not exactly collocated (three of the four are within 10 km of each other, the fourth—the 28–30 June section—was within 25 km), these sections were close to one another and demonstrate significant variability of the overflow in the Western Valley in a period of days or weeks. The mean (standard deviation) overflow transport from the 28 transects is 0.43 Sv (0.48 Sv), (Fig. 8), with a minimum of 0.05 Sv and a maximum of 2.13 Sv. These values are likely an underestimate of the total IFR overflow considering that the overflow across the eastern IFR has most likely descended below 1000 m by the time it reaches the Iceland shelf.

Fig. 16.
Fig. 16.

Four representative SG sections intersecting the Iceland shelf in September 2007 and in June 2007, 2008, and 2009. Contours of σΘ are shown in color for the portion of the sections with density greater than 27.65 kg m−3. Two of the sections encounter large overflow events and two see little overflow.

Citation: Journal of Physical Oceanography 43, 11; 10.1175/JPO-D-13-029.1

R. H. Käse et al. (2013, unpublished manuscript) found a seasonal cycle in WV overflow transport in both observations from an ADCP as well as two numerical models. They found maximum overflow transport in summer with considerable variability on time scales from days to weeks. Of the 28 Seaglider sections, 2 were made in February, 10 in June, and 16 in September. Although this distribution cannot establish a seasonal cycle, the mean transport values in June (0.69 Sv) and September (0.29 Sv) are larger than the mean of the two observations in February (0.2 Sv).

The steady overflow current measurements in the WV led Perkins et al. (1998) to postulate that numerous overflow sources and long transit times insulate the narrow Iceland shelfbreak overflow current from the variability of the IFR overflow events. The present observations suggest considerably more variability in the transport of this current. The watershed of overflow in the WV appears to be only the northwestern portion of the IFR, restricting the number of overflow locations and therefore the possibility that long transit times and multiple sources can buffer the Iceland shelfbreak current against variability in its sources.

7. Conclusions

We have presented analysis of three years of new hydrographic data in the region of the eastern Greenland–Scotland Ridge. Mean hydrographic properties support existing knowledge about the Faroe Bank Channel overflow while the large quantity and high spatial resolution of the observations allows for a more detailed description of the region, in particular the distribution of the overflow downstream of the FBC. New estimates of an Iceland–Faroe Ridge overflow in the Western Valley have been presented.

We find geostrophically balanced flow at section I in the FBC with a long-term-mean transport of 1.8 ± 0.2 Sv of dense water (σΘ ≥ 27.8 kg m−3), consistent with measurements at the channel sill (Hansen and Østerhus 2007). The FBC overflow remains thick (≥200 m) until a secondary sill is crossed approximately 50 km downstream of the main sill, at which point the plume thins (~115 m) as it evolves from a channel flow to a gravity current on a slope. This secondary sill coincides with the largest changes in the properties of the overflow plume. Between the second sill and a topographic feature near 10°W, composite sections suggest that the plume is not in geostrophic balance. This region is one where elevated mixing, critical Froude numbers, and hydraulic criticality have been observed in past studies (Fer et al. 2010; Girton et al. 2006; Pratt et al. 2007). Between the secondary sill and the topographic feature, the overflow bifurcates with the densest portion of the plume, and 67% of the total transport, descending below 1000 m. A shallower branch of approximately 0.6 Sv remains at intermediate depths on the IFR. Seaglider-inferred dissipation is also highest in this region (Beaird et al. 2012). We conclude that the region at and just downstream of the second sill of the FBC marks a more important dynamical transition than the primary FBC sill.

On the Atlantic flank of the Iceland–Faroe Ridge, the alongslope overflow transport remains fairly constant at 0.6 Sv for 150 km. Absolute geostrophic velocities above the frictional bottom boundary layer average 21.5 cm s−1, predominantly along bathymetric contours. Rough calculations suggest that the observed-mean alongslope flow must be maintained by approximately 0.8 Sv of overflow from the IFR.

Bottom mixed layers range in thickness from 0 to 70 m with temperatures between 1° and 5°C. Individual profiles show that a near-bottom pycnocline, evidence of overflow, is ubiquitous on the AIFR. This pycnocline is nearly parallel to the slope of the ridge bathymetry, supporting the alongslope-mean circulation. Pycnocline slopes associated with mesoscale variability are a fraction of the cross-ridge slopes, suggesting that the baroclinic velocities of the mesoscale features play a second-order role in the circulation of the overflow plume on the Atlantic flank of the IFR.

Around 13°W, the IFR becomes meridionally aligned and alongslope transport through the sections diminishes to about 0.2 Sv. This portion of the IFR is characterized by relatively warm near-bottom temperatures (mean: 3°–5°C, minimum: 2°–3°C) that isolate the colder and less-diluted overflows on the eastern half of the ridge from the cold overflow waters found in the Western Valley at the Iceland shelf break.

We frequently find relatively undiluted overflow water adjacent to the Iceland shelf in the WV, waters that evidently crossed the IFR, if not through the WV itself, somewhere on the northwestern portion of the IFR. Repeated Seaglider sections normal to the Iceland shelf show that the standard deviation in the overflow transport (0.48 Sv) is as large as the mean (0.43 Sv) in the current adjacent to the Iceland shelf. It appears that significant variability in the overflow transport in the WV exists on time scales from days to weeks.

Acknowledgments

The authors thank Kirk O’Donnell, Bill Fredericks, and James Bennett for help with data collection and processing as well as many enlightening conversations. Also, we thank Mark Prater and Tom Rossby for kindly providing RAFOS float dataand, finally, the crew of the R.V. Magnus Heinason, Bogi Hansen, and Hjálmar Hátún for considerable help from the Faroe Islands. This work has been generously supported by National Science Foundation, OCE Division, through Grants OCE-1029344 and OCE-0550584.

REFERENCES

  • Allen, J., , D. Smeed, , and A. Chadwick, 1994: Eddies and mixing at the Iceland–Faroes Front. Deep-Sea Res., 41, 5179.

  • Beaird, N. L., , I. Fer, , P. B. Rhines, , and C. C. Eriksen, 2012: Dissipation of turbulent kinetic energy inferred from Seagliders: An application to the eastern Nordic seas overflows. J. Phys. Oceanogr., 42, 22682282.

    • Search Google Scholar
    • Export Citation
  • Bretherton, F. P., , R. E. Davis, , and C. B. Fandry, 1976: A technique for objective analysis and design of oceanographic experiments applied to MODE-73. Deep-Sea Res., 23, 559582.

    • Search Google Scholar
    • Export Citation
  • Cenedese, C., , J. A. Whitehead, , T. A. Ascarelli, , and M. Ohiwa, 2004: A dense current flowing down a sloping bottom in a rotating fluid. J. Phys. Oceanogr., 34, 188203.

    • Search Google Scholar
    • Export Citation
  • Darelius, E., , I. Fer, , and D. Quadfasel, 2011: Faroe Bank Channel overflow: Mesoscale variability. J. Phys. Oceanogr., 41, 21372154.

  • Duncan, L. M., , H. L. Bryden, , and S. A. Cunningham, 2003: Friction and mixing in the Faroe Bank Channel outflow. Oceanol. Acta, 26, 473486.

    • Search Google Scholar
    • Export Citation
  • Enmar, L., , K. Borenäs, , I. Lake, , and P. Lundberg, 2009: Comments on “Is the Faroe Bank Channel overflow hydraulically controlled?” J. Phys. Oceanogr., 39, 15341538.

    • Search Google Scholar
    • Export Citation
  • Eriksen, C. C., , T. J. Osse, , R. D. Light, , T. Wen, , T. W. Lehman, , P. L. Sabin, , J. W. Ballard, , and A. M. Chiodi, 2001: Seaglider: A long-range autonomous underwater vehicle for oceanographic research. IEEE J. Oceanic Eng., 26, 424436.

    • Search Google Scholar
    • Export Citation
  • Fer, I., , G. Voet, , K. S. Seim, , B. Rudels, , and K. Latarius, 2010: Intense mixing of the Faroe Bank Channel overflow. Geophys. Res. Lett., 37, L02604, doi:10.1029/2009GL041924.

    • Search Google Scholar
    • Export Citation
  • Fogelqvist, E., , J. Blindheim, , T. Tanhua, , S. Østerhus, , E. Buch, , and F. Rey, 2003: Greenland–Scotland overflow studied by hydro-chemical multivariate analysis. Deep-Sea Res. I, 50, 73102.

    • Search Google Scholar
    • Export Citation
  • Geyer, F., , S. Østerhus, , B. Hansen, , and D. Quadfasel, 2006: Observations of highly regular oscillations in the overflow plume downstream of the Faroe Bank Channel. J. Geophys. Res., 111, C12020, doi:10.1029/2006JC003693.

    • Search Google Scholar
    • Export Citation
  • Gill, A. E., 1982: Atmosphere–Ocean Dynamics. Academic Press, 662 pp.

  • Girton, J. B., , L. J. Pratt, , D. A. Sutherland, , and J. F. Price, 2006: Is the Faroe Bank Channel overflow hydraulically controlled? J. Phys. Oceanogr., 36, 23402349.

    • Search Google Scholar
    • Export Citation
  • Hansen, B., , and J. Meincke, 1979: Eddies and meanders in the Iceland–Faroe Ridge area. Deep-Sea Res., 26, 10671082.

  • Hansen, B., , and S. Østerhus, 2000: North Atlantic–Nordic seas exchanges. Prog. Oceanogr., 45, 109–208.

  • Hansen, B., , and S. Østerhus, 2007: Faroe Bank Channel overflow 1995–2005. Prog. Oceanogr., 75, 817856.

  • Harvey, J. G., , and A. Theodorou, 1986: The circulation of Norwegian Sea overflow water in the eastern North Atlantic. Oceanol. Acta, 9, 393402.

    • Search Google Scholar
    • Export Citation
  • Hermann, F., 1967: The T-S diagram analysis of the water masses over the Iceland–Faroe Ridge and in the Faroe Bank Channel (Overflow ’60). Rapp. P.-V. Reun. - Cons. Int. Explor. Mer,157, 139–149.

  • Johnson, G. C., 2008: Quantifying Antarctic Bottom Water and North Atlantic Deep Water volumes. J. Geophys. Res.,113, C05027, doi:10.1029/2007JC004477.

  • Johnson, G. C., , and T. B. Sanford, 1992: Secondary circulation in the Faroe Bank Channel outflow. J. Phys. Oceanogr., 22, 927933.

  • Killworth, P., 2001: On the rate of descent of overflows. J. Geophys. Res., 106 (C10), 22 26722 275.

  • Mauritzen, C., , J. Price, , T. Sanford, , and D. Torres, 2005: Circulation and mixing in the Faroese Channels. Deep-Sea Res., 52, 883913.

  • Meincke, J., 1972: The hydrographic section along the Iceland–Faroe Ridge carried out by R.V. ‘Anton Dohrn’ in 1959–1971. Ber. Dtsch. Wissenchaftlichen Komm. Meeresforschung, 22, 372384.

    • Search Google Scholar
    • Export Citation
  • Nof, D., , N. Paldor, , and S. V. Gorder, 2002: The reddy maker. Deep-Sea Res., 49, 15311549.

  • Perkins, H., , T. J. Sherwin, , and T. Hopkins, 1994: Amplification of tidal currents by overflow on the Iceland–Faroe Ridge. J. Phys. Oceanogr., 24, 721735.

    • Search Google Scholar
    • Export Citation
  • Perkins, H., , T. S. Hopkins, , S. A. Malmberg, , P. M. Poulain, , and A. Warn-Varnas, 1998: Oceanographic conditions east of Iceland. J. Geophys. Res., 103 (C10), 21 53121 542.

    • Search Google Scholar
    • Export Citation
  • Prater, M. D., , and T. Rossby, 2005: Observations of the Faroe Bank Channel outflow using bottom-following RAFOS floats. Deep-Sea Res., 52, 481494.

    • Search Google Scholar
    • Export Citation
  • Pratt, L. J., , U. Riemenschneider, , and K. R. Helfrich, 2007: A transverse hydraulic jump in a model of the Faroe Bank Channel outflow. Ocean Modell., 19, 19.

    • Search Google Scholar
    • Export Citation
  • Price, J. F., , and M. O. Baringer, 1994: Overflows and deep water production by marginal seas. Prog. Oceanogr., 33, 161200.

  • Riemenschneider, U., , and S. Legg, 2007: Regional simulations of the Faroe Bank Channel overflow in a level model. Ocean Modell., 17, 93122.

    • Search Google Scholar
    • Export Citation
  • Saunders, P. M., 1990: Cold outflow from the Faroe Bank Channel. J. Phys. Oceanogr., 20, 2943.

  • Saunders, P. M., 1996: The flux of dense cold overflow water southeast of Iceland. J. Phys. Oceanogr., 26, 8595.

  • Seim, K. S., , and I. Fer, 2011: Mixing in the stratified interface of the Faroe Bank Channel overflow: The role of transverse circulation and internal waves. J. Geophys. Res., 116, C07022, doi:10.1029/2010JC006805.

    • Search Google Scholar
    • Export Citation
  • Seim, K. S., , I. Fer, , and J. Berntsen, 2010: Regional simulations of the Faroe Bank Channel overflow using a σ-coordinate ocean model. Ocean Modell., 35, 3144.

    • Search Google Scholar
    • Export Citation
  • Steele, J. H., 1959: Observations of deep water overflow across the Iceland–Faroe Ridge. Deep-Sea Res., 6, 7072.

  • Swift, J., 1984: The circulation of the Denmark Strait and Iceland–Scotland overflow waters in the North Atlantic. Deep-Sea Res., 31, 13391355.

    • Search Google Scholar
    • Export Citation
  • Wilkenskjeld, S., , and D. Quadfasel, 2005: Response of the Greenland–Scotland overflow to changing deep water supply from the Arctic Mediterranean. Geophys. Res. Lett., 32, L21607, doi:10.1029/2005GL024140.

    • Search Google Scholar
    • Export Citation
  • Xu, X., , W. Schmitz, , H. Hurlburt, , P. Hogan, , and E. Chassignet, 2010: Transport of Nordic seas overflow water into and within the Irminger Sea: An eddy-resolving simulation and observations. Geophys. Res. Lett., 115, C12048, doi:10.1029/2010JC006351.

    • Search Google Scholar
    • Export Citation
Save