Discovery of an unrecognized pathway carrying overflow waters toward the Faroe Bank Channel

The Atlantic Meridional Overturning Circulation (AMOC) transports warm, saline waters northward that cool, sink, and return as dense overflow waters. A key component is the Faroe Bank Channel Overflow (FBCO), one of two main conduits (with Denmark Strait) carrying dense water from the Nordic Seas into the North Atlantic. Long-term ADCP observations show the FBCO contributes about one-third (~2 Sv) of the total overflow (~5.8 Sv) and has remained stable without long-term slowdown. Despite its importance for AMOC stability, the specific source regions and variable pathways feeding the FBCO are not well constrained. Many studies have presumed the overflow is supplied primarily by deep waters approaching from the western Nordic Seas and entering the Faroe-Shetland Channel (FSC) directly north of the Faroes. Modeling work, however, has hinted at an additional eastern pathway from the Norwegian slope that may vary with wind forcing. This study aims to identify and document the existence, origin, and mechanisms of an eastern overflow path and to determine how overflow waters are routed within the FSC, using observations and an eddy-resolving ocean model.

Earlier observational and conceptual work established the role and stability of Nordic Seas overflows in the AMOC (e.g., Dickson and Brown 1994; Mauritzen 1996; Hansen & Østerhus 2000, 2007; Olsen et al. 2008; Østerhus et al. 2019). Many observational studies assumed the FBCO is primarily fed by deep waters entering the FSC from the west, passing north of the Faroes (e.g., Søiland et al. 2008). Float-based analyses suggested waters within the 1750 m contour northwest of the Faroe plateau could be routed directly under the Faroe Current into the FBCO. In contrast, modeling studies indicated two branches: a western/direct path and an eastern/indirect path along the Norwegian slope or Jan Mayen Ridge, with relative strengths modulated by wind forcing (Serra et al. 2010; Köhl 2010). Other studies explored wind and thermohaline controls on overflow variability, and large-scale Nordic Seas exchanges (e.g., Biastoch et al. 2003; Sandø et al. 2012; Bringedal et al. 2018). However, direct observational evidence for an eastern pathway from the Norwegian slope feeding the FBCO and clarification of which FSC boundary predominantly carries the overflow remained lacking.

The study combines multi-platform observations with an eddy-resolving ocean model and Lagrangian analysis.

• FBCO observations: Since Nov 1995, a bottom-mounted upward-looking 75 kHz RDI BroadBand ADCP near the sill in the Faroe Bank Channel provides a continuous kinematic overflow time series representative of FBCO transport (except annual servicing breaks).
• Deep flow under the Faroe Current (FCdeep): An upward-looking 75 kHz ADCP at 62.92°N, 6.08°W (bottom depth ~950 m) records daily averaged horizontal velocities at 628 m depth beneath the Faroe Current’s Atlantic layer (continuous since July 1998 aside from servicing gaps).
• Vessel-mounted ADCP across FSC: A 75 kHz RDI Ocean Surveyor ADCP on MS Norröna (2008–2016) provided transects across the FSC, operating in narrow-band mode, reaching ~600 m depth (effective returns to ~500 m). Profiles averaged every 3 minutes (~3 km along track) with 16 m vertical resolution and ±0.02 m s−1 uncertainty; headings from a Thales ADU-5 (accuracy 0.03°). Data were detided.
• Moored ADCPs across FSC: S-line moorings (SG/SC/SY/SB) spanning depths ~786–1069 m with deepest bins at ~617–810 m used to resolve deep boundary currents on both sides of the channel.
• Atmospheric reanalysis: ERA-Interim sea level pressure (daily and monthly; 1°×1°; 1993–2016).
• Satellite altimetry: DUACS DT2014 multimission daily SSH (0.25°×0.25°; Jan 1993–Apr 2016) to assess basin-scale sea-level and barotropic signals.
• Ocean model: NEMO v3.6 global eddy-resolving simulation ORCA0083-N001 (nominal 1/12°; 75 z* levels), forced by DRAKKAR forcing set v5.2 (ERA-40 then ERA-Interim), free-running with SSS restoring; spun up from 1958, analysis from 1979–2010.
• Lagrangian trajectories: TRACMASS v6.0 used to backtrack numerical particles seeded every five days in FSC grid cells carrying overflow. Particles traced backward up to five years, until reaching ~65°N domain boundary or exiting; only trajectories with temperature <2 °C retained for deep overflow characterization. Arrival-time sensitivity tested; statistics computed on those reaching the northern boundary within one year (others excluded).
• Statistical analyses: Time series detrended and smoothed with a 360-day running mean for interannual analyses; correlations and composite differences computed. Significance assessed using a random phase test with 20,000 Monte Carlo simulations. Seasonal cycles computed from monthly means.
• Volume flux calculations: Modeled along-channel transport ψ(t)=∫₀ᴸ∫₀ʰ(x) v(x,z,t) dz dx, computed separately for western and eastern halves of FSC divided by the channel thalweg; layer-specific fluxes below 482 m also evaluated (CDFTOOLS).

• Identification of an unrecognized eastern pathway: Observations and model results show that dense overflow waters feeding the FBCO can follow an indirect path along the Norwegian slope before turning south into the FSC, in addition to the traditionally assumed direct western approach.
• Atmospheric control: Anticyclonic anomalies over the Nordic Seas strengthen the FBCO and simultaneously weaken or reverse the eastward FCdeep; cyclonic anomalies have the opposite effect. Composite SLP and SSH patterns (altimetry) indicate a wind-driven barotropic regulation along closed f/H contours. On seasonal scales, winter (stronger cyclonic winds) corresponds to lower FBCO and stronger FCdeep; summer shows higher FBCO and weaker or reversed FCdeep.
• Strong FCdeep–FBCO anticorrelation: Interannual and seasonal time series exhibit an inverse relationship between FCdeep and FBCO (r = −0.69, p < 0.01; 1999–2016; after detrending and 360-day smoothing), inconsistent with a solely direct western approach and implying an alternating eastern path.
• Pathway bimodality from Lagrangian analysis: Backward trajectories of T<2 °C waters reveal two distinct routes. Early 1990s (anomalously cyclonic) show dominance of the western/direct path and weaker FBCO; early 2000s (anomalously anticyclonic) show strong activation of the eastern/indirect path along the Norwegian slope and stronger FBCO. Probability distributions of end longitudes are unimodal (western) under cyclonic forcing and bimodal under anticyclonic forcing.
• Discovery of the Faroe-Shetland Channel Jet (FSCJ): Both model and observations reveal a persistent, bottom-intensified deep jet banked against the eastern (Shetland) slope of the FSC. This jet carries the bulk of the overflow transport through the channel, contrary to the previous assumption of western-boundary dominance. The jet also exhibits an intermediate-depth core.
• Transport partitioning: Modeled volume transports show the eastern boundary (FSCJ) dominates the deep overflow flux in a time-mean sense and during high-overflow periods (early 2000s). Moored ADCPs across the FSC corroborate strongest southward deep velocities at the eastern boundary.
• Water mass properties: Particles reaching the FSC via the indirect eastern path are colder than those arriving via the direct western route. The modeled FSCJ transport correlates positively with deep density on the eastern side of the channel, indicating that enhanced FSCJ coincides with denser overflow waters.
• Source depth dependence: Under cyclonic forcing, flows are constrained to shallower isobaths connecting directly to the FSC; under anticyclonic forcing, a deeper route opens, likely tapping waters from along the Jan Mayen Ridge, enhancing the eastern path contribution.

The study resolves a long-standing uncertainty about the upstream routing of overflow waters feeding the FBCO. While the general Nordic Seas circulation is cyclonic and source waters must pass north of the Faroes, the specific approach to the FSC alternates between a short western/direct route and a longer eastern/indirect route along the Norwegian slope, depending on the large-scale wind forcing. Anticyclonic atmospheric regimes weaken the basin cyclonic circulation, promote a barotropic adjustment evident in SSH, and activate the eastern path, increasing FBCO transport. Cyclonic regimes favor the direct western path and reduced FBCO. Regardless of upstream approach, the overflow is predominantly carried by a deep, bottom-intensified Faroe-Shetland Channel Jet along the eastern boundary, redefining the canonical view of FSC dynamics. The FSCJ’s strengthening is associated with denser overflow waters, implying pathway-dependent water mass properties and potential impacts on downstream subpolar overturning. The FSCJ’s structure and role bear resemblance to the North Icelandic Jet feeding Denmark Strait overflow, suggesting analogous boundary-intensified dynamics. These findings link atmospheric variability to overflow routing and intensity via basin-scale, largely barotropic, circulation changes and topographic steering, providing a framework to interpret observed variability in FBCO and FCdeep.

This work presents observational and modeling evidence for a previously unrecognized eastern pathway along the Norwegian slope that feeds the Faroe Bank Channel Overflow and demonstrates that the primary conduit through the Faroe-Shetland Channel is a deep, bottom-intensified jet along the eastern boundary (FSCJ). Anticyclonic wind forcing over the Nordic Seas activates the eastern pathway and strengthens the overflow, while cyclonic forcing does the opposite and favors the direct western route. The FSCJ carries most of the overflow transport independent of the upstream approach and is associated with denser overflow waters when enhanced. These results revise the conceptual model of overflow routing into the North Atlantic and clarify links between atmospheric forcing, basin circulation, and overflow variability, with implications for the stability of the lower branch of the AMOC. Future work should quantify the dynamics and stability of the FSCJ, its sensitivity to changing wind and buoyancy forcing under climate change, and the downstream impacts of pathway-dependent water mass properties.

• Observational coverage: The vessel-mounted ADCP resolves velocities only to ~500–600 m depth, requiring moorings to infer deeper structure; moored arrays sample limited cross-channel locations. Data gaps occur during servicing periods.
• Modeling constraints: The Lagrangian pathway analysis relies on a single eddy-resolving model realization with inherent biases; it is free-running with SSS restoring and limited to 1979–2010 for analysis. Pathway identification depends on temperature thresholding (T<2 °C) and seeding/termination criteria.
• Statistical assumptions: Interannual analyses use detrending and heavy smoothing (360-day running mean); significance is assessed via random phase tests, which assume specific spectral properties.
• Temporal variability: The strong FCdeep–FBCO anticorrelation weakens after ~2008 on interannual scales, suggesting additional processes or nonstationarity not fully resolved here.
• Mechanistic detail: While the FSCJ is documented, its detailed dynamics, topographic controls, and sensitivity to forcing changes are not fully explored and warrant further dedicated study.