Circum-Antarctic bottom water formation mediated by tides and topographic waves

Antarctic Bottom Water (AABW) forms from dense shelf water (DSW) created by brine rejection during sea-ice growth and ocean/ice-shelf interactions. DSW overflows the shelf break and descends the continental slope, entraining ambient waters to produce AABW, ventilating the abyss and affecting global heat and carbon budgets. Historically, the Weddell and Ross Seas were considered primary AABW sources, with Adélie Coast and Prydz Bay also identified recently. Observations show dense overflows are widespread around Antarctica. Theoretically, rotational effects turn steady downslope flows along isobaths with slow Ekman descent; variable bathymetry can aid descent via topographic steering. However, observations show DSW reaching deep ocean over short along-slope distances even without steering, implying additional mechanisms accelerate descent. Two candidates are tides and overflow-forced topographic Rossby waves (TRWs). The study synthesizes historical observations (Weddell and Ross Seas) and uses idealized, high-resolution numerical experiments to determine how tides and TRWs, separately and jointly, control DSW downslope transport, entrainment, AABW formation rates, and properties across Antarctic overflow regimes.

• Source regions of AABW: Weddell and Ross Seas as primary sources (Gill 1973; Orsi et al. 1999), with additional formation along Adélie Coast and Prydz Bay (Rintoul 1985; Williams et al. 2008; Ohshima et al. 2013). Dense overflows are common around the Antarctic margin (Amblas & Dowdeswell 2018).
• Theoretical background: Rotational dynamics and benthic Ekman controls on descent rates (Killworth 2001); bathymetric steering (Darelius & Wåhlin 2007; Wang et al. 2009). Overflow instabilities and TRWs (Swaters 1991; Poulin & Swaters 1999; Stewart & Thompson 2016; Han et al. 2022, 2023).
• Tidal influences: Observations and modeling indicate tides regulate overflow strength and properties in the Ross Sea (Whitworth & Orsi 2006; Padman et al. 2009; Guan et al. 2009; Wang et al. 2010; Bowen et al. 2021, 2023). Mixing efficiencies and implications for entrainment (Gregg et al. 2018). Tidal impacts on Antarctic Slope Front/Current (Flexas et al. 2015; Stewart et al. 2019).
• Observational characteristics in Weddell Sea: Low-frequency variability and TRWs on the slope (Darelius et al. 2009; Jensen et al. 2013; Gordon et al. 2020; Semper & Darelius 2017). Ice-shelf water properties and circulation context (Nicholls et al. 2009; Janout et al. 2021; Thompson et al. 2018).
• Modeling practices: Use of ROMS and turbulence closure, overflow parameterization, and need for improved representation in climate models (Legg et al. 2009; Heuzé 2021; Shchepetkin & McWilliams 2003; Mellor & Yamada 1982; Ilicak et al. 2011).

Observations: Compiled historical in situ moorings and hydrography from Weddell Sea (19 moorings, 1968–2011, each ~1–2 years) and Ross Sea (AnSlope 2003–2005; CALM 2007–2011) targeting overflow sites. Used hourly near-bottom instruments (~20 mab Ross; 25–125 mab Weddell) to minimize blow-down effects. Employed CTD sections in Ross Sea to analyze cross-slope hydrography. For validation, used Adélie coastal region mooring (Mertz Polynya Experiments) with ~500 days of ADCP velocity at ~1162 m (~20 mab) and nearby temperature sensors.

Modeling: Employed Regional Ocean Modeling System (ROMS) for process-oriented, eddy-resolving simulations. Domain ~1200 km × 650 km with an embayment connected to a flat 2500 m abyss via a linear continental slope; included a western trough. Horizontal grid stretched (0.5 km near trough to ~4 km at open boundaries); 60 terrain-following vertical levels with enhanced near-bottom resolution (~5 m over upper slope). Constant f-plane at 72°S (f = 1.38×10^-4 s^-1). Vertical mixing via Mellor–Yamada 2.5 scheme; quadratic bottom drag Cd = 0.003. Initial stratification adapted from Ross Sea observations (strong upper ~200 m stratification, weaker below).

Experiments: Conducted Ross Sea-like simulations with trough width 30 km and slope s = ΔH/Δx = 0.15; restored DSW at the southern trough boundary to produce ~0.6 Sv shelf-break overflow (ρ > 1027.86 kg/m³). Weddell Sea-like setup with trough width 50 km; slope varying from ~0.07 (upper slope) to ~0.02 (lower slope); restored ice-shelf water-like DSW (−2.2 °C, 34.73 psu), generating ~0.7 Sv overflow (smaller than observed ~1.6 ± 0.5 Sv). Sensitivity experiments varied slope steepness across Antarctic-relevant regimes, with smaller DSW flux (~0.3 Sv) to map behaviors.

Tidal forcing: Imposed K1 tidal sea surface height at open boundaries extracted from TPXO7 near Ross Sea. Ran base integrations 60–100 days without tides, then branched 30 days with/without tides; analyzed last 10 days (quasi-steady overflow fluxes). Computed daily tidal flow strength S_t = ⟨u(t)v(t)⟩ (daily moving average). Maximum tidal flow V_tide from vertically averaged (upper 200 m) daily maxima over 10 days. Estimated diurnal tidal excursion L using u_m (representative cross-slope speed) and period T.

Diagnostics: Tracked DSW descent via passive tracer injected with DSW. Computed isobath-weighted tracer center-of-mass depth/isobath H_DSW(x) = ∫ τ H dz dy / ∫ τ dz dy (10-day averages) to represent mean descent pathway. Assessed diapycnal tracer mass flux across a density surface (e.g., 27.88 kg/m³ referenced to surface) via zonally integrated Hovmöller diagrams. Performed spectral analysis (velocity PSD) at selected slope locations to identify periodicities (diurnal tides, multi-day TRWs). Classified circum-Antarctic regimes using slope steepness (RTopo-2, 30 arcsec) between 800–1800 m and summed major tidal constituent speeds (O1, K1, M2, S2) to infer V_tide. Compared modeled hydrographic structures to observed CTD sections; constructed T–S diagrams below 1500 m for AABW properties.

• Distinct variability at major overflow sites: Ross Sea shows strong diurnal oscillations consistent with tides; Weddell Sea shows multi-day oscillations consistent with topographic Rossby waves (TRWs). At Ross slope mooring, correlation between daily tidal flow strength and minimum temperature is −0.8, with colder waters during spring tides, indicating tidal advection of DSW dominates temperature variability on slope.
• Tidal impacts (Ross-like regime): Including K1 tides advects DSW tracers to greater depths and maintains higher tracer concentrations at depth, while increasing dilution in shallow regions (<500 m). Diapycnal tracer mass flux across a dense isopycnal is larger and more temporally steady without tides; with tides it is weaker and periodic, showing overall reduced cumulative entrainment during descent. Tides deepen the pycnocline episodically (notably during ebb), creating a V-shaped front that drives cold, fresher surface waters to >1000 m and couples them to DSW, yielding colder, fresher AABW in T–S space relative to no-tide cases and to observations.
• Weddell-like dynamics: Cross-slope tidal speeds at 1000 m isobath are relatively weak; taking u_m ≈ 20 cm s−1 gives diurnal tidal excursion ~5 km and vertical displacements <250 m, insufficient to drive efficient downslope transport. TRWs dominate with higher-frequency signals on upper slope and lower-frequency on lower slope; TRW-associated offshore phases coincide with DSW downslope transport. TRWs form isolated eddies on upper slope (shallower than ~1500 m) that temporarily confine DSW boluses. TRWs alone can generate V-shaped fronts and likely reduce entrainment, producing colder/fresher AABW, similar to tides.
• Combined effects and slope dependence: Without tides, DSW descent rate varies non-monotonically with slope steepness due to transitions among steady geostrophic flows (steep), TRW-dominated flows (moderate), and eddying (gentle). With tides, DSW reaches deep ocean more rapidly for steep slopes (s* ≥ ~0.125), whereas for gentler slopes (s* ≤ ~0.1) tides have little effect on the mean descent pathway because TRWs govern downslope transport. Thus, tides and TRWs play comparable roles, but tidal influence is significant mainly where steep slopes suppress TRWs.
• Dynamical regime classification: Four regimes identified by slope and tides: Mixing (s > 0.1, V_tide < 20 cm s−1) with strong entrainment and slow descent; Tidal (s > 0.1, V_tide ≥ 20 cm s−1) with accelerated descent and reduced entrainment; Wavy (0.05 ≤ s ≤ 0.1) with TRW-accelerated descent and reduced entrainment; Eddying (s < 0.05) with isolated eddies on upper slope delaying descent. Most Antarctic margins fall outside the Mixing regime, implying generally favorable conditions for AABW formation where DSW is produced.
• Regional validation and mapping: Circum-Antarctic map combining slope steepness (RTopo-2) and tidal current speeds (sum of O1, K1, M2, S2) predicts dominant regimes around the margin. Independent moorings confirm tidal dominance in Adélie (diurnal and fortnightly signals; colder water during ebb/spring tides; no subtidal TRWs) and TRW dominance in Prydz Bay (subtidal wavy signals). Polynya occurrence (2013–2021) suggests many additional shelf regions may generate DSW and feed AABW under favorable regimes, particularly in East Antarctica.
• Sensitivity to DSW density: Varying DSW density shows minimal impact on overflow dynamics for moderate/small slopes; for steep slopes, tidal enhancement of AABW density is less pronounced for lighter DSW, potentially due to weaker interfacial stratification enhancing shear-driven mixing under tides. Overall results robust across a range of DSW properties.
• Quantitative details: Ross Sea DSW flux in model ~0.6 Sv (observed ~0.8 Sv); Weddell model flux ~0.7 Sv (observed ~1.6 ± 0.5 Sv). Representative Weddell tidal excursion ~5 km for u_m = 20 cm s−1; vertical displacement <250 m. Observed Ross slope temperature–tide correlation −0.8.

The study resolves how tides and TRWs control DSW descent and entrainment, thereby shaping AABW formation rates and properties. Both mechanisms accelerate downslope transport, reducing cumulative entrainment and producing colder, denser (and fresher) AABW relative to cases dominated by steady geostrophic descent and mixing. However, their relative importance depends strongly on continental slope steepness: on steep slopes, TRWs are suppressed, allowing tides to dominate and substantially impact descent; on gentler slopes, TRWs provide the primary acceleration and tides contribute little. Both processes generate V-shaped hydrographic fronts at the shelf break, facilitating contact between DSW and cold, fresh surface waters, implying that surface water changes can directly influence abyssal water properties. The derived four-regime framework and circum-Antarctic classification provide predictive guidance on where AABW formation is favored and inform observational strategies. Because high-frequency tides and TRWs are not routinely resolved or parameterized in climate models, these findings highlight a need to incorporate their effects to improve simulations of the global overturning and abyssal ventilation, with implications for heat/carbon sequestration and millennial-scale climate variability.

By synthesizing multi-decadal observations and conducting high-resolution, process-oriented simulations, the study demonstrates that tides and topographic Rossby waves play comparable roles in mediating DSW descent and AABW formation. Both processes accelerate downslope transport and reduce entrainment, leading to colder, denser, and fresher AABW. The continental slope steepness sets regimes: tides dominate on steep slopes where TRWs are suppressed, while TRWs dominate on gentler slopes. A four-regime classification (Mixing, Tidal, Wavy, Eddying) and a circum-Antarctic map identify regions favorable for AABW formation and are supported by independent moorings (Adélie, Prydz). These results emphasize the pervasive role of high-frequency processes that are poorly represented in current climate models. Future work should incorporate ASF/ASC dynamics, winds, freshwater inputs, and surface water changes into more realistic simulations; better constrain minimal spatial scales for regime development in variable bathymetry; and further disentangle tidal advection versus mixing impacts on overflow transformation and biogeochemical transport.

• Idealized modeling: Process-oriented ROMS configurations exclude atmospheric forcing, winds, freshwater input, and interactions with the Antarctic Slope Front/Current (ASF/ASC), limiting realism.
• Temporal scope: Analyses focus on last 10 days of branched 30-day with/without tide experiments; full statistical equilibrium is not expected.
• TRW control experiment: In Weddell-like setup, TRWs are internally generated and cannot be selectively removed, precluding direct with/without TRW comparisons.
• Tidal forcing simplification: Boundary tides include K1 constituent for main process analysis; tidal representation is simplified and domain scaled relative to Ross Sea.
• Parameter sensitivity: Regime classification is sensitive to local slope steepness and along-slope bathymetric variability, which may limit development of Wavy/Eddying regimes; mapping resolution and thresholds (e.g., V_tide ≥ 20 cm s−1) are approximations.
• Observational constraints: Mooring datasets differ in sampling and sensor depth; near-bottom focus avoids blow-down but reduces vertical coverage; circum-Antarctic overflow detection remains sparse. Presence of polynyas does not guarantee local DSW formation due to shelf stratification/circulation constraints.
• DSW property choices: Restored DSW properties in models (e.g., Weddell denser than observed) and flux magnitudes differ from observations, which may influence quantitative but not qualitative outcomes.