Winter seal-based observations reveal glacial meltwater surfacing in the southeastern Amundsen Sea

Antarctic ice shelves are thinning rapidly due primarily to basal melting driven by the intrusion of warm modified Circumpolar Deep Water (mCDW) into ice-shelf cavities. As mCDW circulates beneath ice shelves, it melts the ice base and produces a relatively fresh, meltwater-rich outflow that is colder than mCDW but warmer than Winter Water (WW). The fate of this meltwater depends on the density at which it reaches neutral buoyancy: if sufficiently buoyant, it can rise toward the surface, potentially offsetting brine rejection from sea-ice formation, reducing deep convection and bottom-water formation, melting sea ice and promoting polynyas, altering air-sea heat fluxes, enhancing overturning circulation, and supplying bioavailable iron. Despite its outsized climatic importance, the spatial pathways and vertical distribution of meltwater remain poorly observed, especially in winter. Summer hydrographic detection of near-surface meltwater is confounded by solar warming of the upper ocean, leading to potential misinterpretation. The Pine Island Bay (PIB) gyre likely influences meltwater pathways, but sparse observations have left this role unclear. This study exploits novel winter seal-tag hydrographic observations to unambiguously map the winter distribution of glacial meltwater near Pine Island Ice Shelf (PIIS), quantify its vertical structure, and assess the role of the PIB gyre in its transport.

Prior work documents rapid thinning and strong basal melt of West Antarctic ice shelves, including PIIS and Thwaites, associated with mCDW intrusions via bathymetric troughs. Observations have identified meltwater in front of PIIS, but near-surface signals in summer are often attributed to solar warming artifacts, shifting focus to a deep meltwater layer (∼200–400 m). Noble gases have been established as robust, unambiguous tracers of glacial meltwater relative to hydrographic tracers affected by air-sea interactions. The cyclonic PIB gyre has been observed and modeled from the surface to ∼700 m and is hypothesized to influence meltwater transport. Sub-ice-shelf basal channels can focus meltwater outflow into buoyant plumes. Modeling studies suggest meltwater can enhance overturning and modulate heat flux into cavities, while observations and models indicate meltwater affects sea ice, polynyas, and iron supply. However, winter distributions and vertical pathways across seasons remained unknown due to observational gaps.

The study combines summer ship-based and winter animal-borne hydrographic datasets and derives meltwater fractions using established tracer methods.

• Ship-based datasets (February 2014, RRS James Clark Ross, iSTAR JR294/295):
  • CTD: Sea-Bird SBE 911 with dual temperature and conductivity sensors; temperature calibrated with SBE 35; salinity with Guildline Autosal; data averaged into 2-dbar bins; TEOS-10 framework applied. Nine CTD stations along the PIIS calving front (median ∼0.5 km from front) on 11 Feb; seven CTD stations along a NW transect (9–20 Feb).
  • VMP-2000: 58 yoyo profiles of temperature and salinity along the calving front (median ∼1 km from front) on 12–13 Feb; salinity calibrated against CTD; data averaged to 0.25 dbar.
  • Noble gases: 71 water samples at CTD stations along the calving front collected in copper tubes, analyzed at WHOI for Ne, Ar, Kr, Xe using cryogenic separation and mass spectrometry; reproducibility 0.1–1.8%, precision ∼0.5–1%; He excluded due to mantle influence.
• Seal-tag hydrographic dataset (winter, July–September 2014):
  • CTD-Satellite Relay Data Loggers deployed on southern elephant and Weddell seals during the cruise; three tagged seals occupied PIB during winter; the four analyzed sections were all sampled by the same female elephant seal (EF838). 625 full-depth winter profiles (temperature, salinity, pressure), with sampling every 2 s onboard the tag, 5 s median filter, compressed to 18 depth points per transmitted profile (including 2 dbar, maximum depth, temperature minimum, deep temperature maximum; remaining points evenly spaced). Only the deepest cast per 4 h transmitted; ascent profiles used. Quality-controlled, linearly interpolated profiles from MEOP; two anomalous salinity points removed. Estimated accuracies: ±0.03 °C (T), ±0.05 g kg−1 (SA).
  • Spatial sections defined: Section 1 along the PIIS calving front; Section 2 from the southeastern end of the front toward the NW across the PIB gyre; Section 3 from the southeastern end westward downstream along the coast with gyre circulation; Section 4 from the northeastern end NW against gyre circulation. Seal-tag data binned horizontally at 200 m along-section intervals for transects.
  • Winter sampling window (1 Jul–10 Sep 2014) ensures absence of solar radiation and presence of a cold, well-mixed WW layer in the upper ocean.
• Meltwater fraction calculations:
  • Hydrographic composite tracer method: Conservative temperature (θ) and Absolute Salinity (SA) used to solve for fractions of mCDW, WW, and glacial meltwater between the mCDW–WW mixing line and the mCDW–meltwater Gade line. Endpoints: WW θ = −1.86 °C, SA = 34.32 g kg−1; mCDW θ = 1.15 °C (summer; winter 1.12 °C tested), SA = 34.87 g kg−1; meltwater θ = −90.8 °C, SA = 0. Calculations repeated with winter mCDW endpoint to assess sensitivity (mean/median meltwater increases of ∼2–5% per section; spatial patterns unchanged). Monte Carlo perturbations (±0.03 °C, ±0.05 g kg−1) indicate ±2.87 g kg−1 uncertainty in meltwater content (∼30% of average winter near-surface value).
  • Noble gas Optimum Multiparameter Analysis (OMP): Fractions of mCDW, air-equilibrated water (AEW), and glacial meltwater inferred from Ne, Ar, Kr, Xe with nonnegativity-constrained least squares; published endpoints used; estimated meltwater uncertainty ±0.5 g kg−1.
• Surface flux effects on derived winter near-surface meltwater: ERA5 net heat flux (∼3×10^7 J m−2 day−1) and climatological salt flux (∼1 kg m−2 day−1) over a 400 m mixed layer imply cooling of ∼0.02 °C day−1 and salinity increase ∼3×10^−3 g kg−1 day−1, equivalent to an apparent decrease of ∼2 g kg−1 day−1 in derived meltwater, representing a lower bound for true near-surface meltwater.
• Hydrographic context: θ–SA diagrams constructed; isopycnals (potential density) used to delineate WW (∼27.45–27.55 kg m−3) and mCDW; isopycnal doming identifies PIB gyre center and structure.

• First unambiguous winter detection of near-surface glacial meltwater in front of Pine Island Ice Shelf (PIIS) using seal-borne hydrography.
• Vertical structure in winter comprises two meltwater-rich layers: a near-surface layer above ∼250 m (above the ∼27.45 isopycnal) and a deep layer near ∼450 m (around the ∼27.55 isopycnal), connected by discrete meltwater-rich columns that traverse the WW layer; large parts of the WW layer remain meltwater-poor.
• Near-surface winter warming signature of ∼0.6 °C above freezing spreads from the northeastern PIIS front southwestward and along the coast toward Thwaites, consistent with the cyclonic PIB gyre pathway; similar features observed near other Amundsen Sea ice shelves.
• Spatial heterogeneity: In contrast to summer’s relatively uniform upper-ocean meltwater content (>5 g kg−1 above the 27.55 isopycnal), winter distributions show strong horizontal and vertical variability with localized columns of high meltwater content.
• PIB gyre control: Isopycnal doming marks the gyre center; both summer and winter sections show highest meltwater content near the ice-shelf front and the gyre’s northwestern edge, with a minimum near the gyre center, indicating cyclonic advection of meltwater.
• Extent of near-surface meltwater in winter differs by section/orientation: detectable to ∼23 km upstream against the gyre (Section 4), ∼35 km across the gyre (Section 2), and at least 70 km downstream along the gyre (Section 3), highlighting gyre influence on transport and dispersion.
• Seasonal mechanism: Winter’s denser, homogeneous upper ocean enhances the relative buoyancy of meltwater, allowing ascent to shallower depths without strong lateral mixing; summer stratification promotes intense lateral mixing of rising meltwater, yielding more uniform distributions.
• Implications: Near-surface meltwater supplies heat capable of maintaining/enhancing polynyas and likely delivers iron to the euphotic zone prior to spring, with potential feedbacks on air-sea heat fluxes, ice-shelf basal melt, iceberg calving, and local overturning.
• Quantitative uncertainties: Hydrographic meltwater estimates in winter have ±2.87 g kg−1 uncertainty from instrument accuracy; surface fluxes can reduce apparent near-surface meltwater by ∼2 g kg−1 day−1, implying observed values are lower bounds.

The study addresses the lack of winter observations of glacial meltwater pathways by demonstrating that meltwater signatures are clearest in winter, when solar warming is absent and the upper ocean is homogenized. The two-layer meltwater structure and connecting columns suggest that both pycnocline trapping (at the perennial mCDW–WW interface) and buoyant plume ascent contribute to the winter distribution. Sub-ice basal channels likely focus outflows into vertical columns that can transit the winter WW layer with limited lateral mixing. The PIB gyre exerts a first-order control on horizontal distribution, concentrating meltwater near the shelf front and gyre periphery and reducing it near the gyre center, with advection pathways reflected in the differing downstream extents among sections. Seasonally, summer stratification increases lateral mixing of rising plumes, producing relatively uniform meltwater in the upper ocean, whereas winter weak stratification preserves heterogeneity. These dynamics imply that wintertime meltwater delivers concentrated heat and freshwater to the surface layer, helping to maintain polynyas and modulating air-sea heat exchange. The associated heat and freshwater fluxes may enhance local ventilation and overturning, potentially increasing heat transport into the cavity and influencing basal melt. The results highlight the need to represent seasonally varying, small-scale meltwater pathways in climate and Earth system models to avoid biases in circulation, ice–ocean interactions, and surface fluxes.

Seal-borne winter hydrographic observations in Pine Island Bay reveal a previously undocumented near-surface meltwater-rich layer coexisting with a deeper meltwater layer and connected by vertical columns, establishing that glacial meltwater can rise to the surface in winter with limited lateral mixing. The Pine Island Bay gyre structures the horizontal distribution, transporting meltwater cyclonically and creating minima at the gyre center. These findings imply that winter meltwater supplies near-surface heat that can maintain and expand polynyas and potentially elevate nutrient (iron) availability before the spring bloom, with cascading impacts on air-sea exchange, local overturning, and ice-shelf basal melt. The work underscores that small-scale, seasonally varying meltwater processes must be captured in models. Future research should obtain multi-year winter observations, including noble gas tracers, to constrain interannual variability, quantify true near-surface meltwater fluxes in the presence of air-sea-ice interactions, and resolve the roles of sub-ice channels and gyre dynamics in setting meltwater pathways.

• Temporal limitation: Only one winter (2014) of seal-tag data in front of PIIS, insufficient to establish long-term mean behavior or interannual variability driven by large-scale climate modes (e.g., ENSO, SAM).
• Tracer endpoint sensitivity: Hydrographic meltwater fractions depend on chosen water-mass endpoints; while tested seasonal variations in mCDW endpoints changed magnitudes modestly, patterns remained robust.
• Measurement uncertainties: Seal-tag hydrographic accuracy (±0.03 °C, ±0.05 g kg−1) induces ±2.87 g kg−1 uncertainty in meltwater estimates; surface flux datasets (heat/salt) used to assess near-surface erosion carry uncertainties.
• Near-surface bias: Winter air-sea-ice processes (cooling, brine rejection) can rapidly erode temperature and salinity signatures, making derived near-surface meltwater fractions a lower bound on true meltwater presence and flux.
• Noble gas coverage: No winter noble gas measurements were available; summer noble gases were used to validate hydrographic methods in stratified conditions only.
• Assumptions about WW properties: Endpoints for summertime WW were based on preceding winter’s WW (2013), assuming full replenishment; interannual variability may affect applicability.