Synergistic atmosphere-ocean-ice influences have driven the 2023 all-time Antarctic sea-ice record low

Antarctic sea ice exerts major influence on the global climate system through air–sea exchanges, ocean circulation, ice-sheet stability and surface mass balance, and polar ecosystems. After a slight long-term increase during 1979–2014, Antarctic SIE has exhibited frequent record lows since 2016. On 21 February 2023, SIE reached a new summer minimum of 1.79 million km² (6.8% lower than the 2022 minimum and ~30% below the 1981–2010 climatology), and negative anomalies persisted through the 2023 advance season, culminating in the lowest winter maximum on 7 September 2023 (16.98 million km²). To understand the mechanisms behind these extremes, this study investigates how atmosphere–ocean–ice processes over March 2022–February 2023 set the stage for the 2023 minimum, with a particular focus on the role of the Amundsen Sea Low (ASL), heat advection, ocean subsurface conditions, and ice–albedo feedbacks.

Prior work on the 2016–2017 spring–summer Antarctic sea-ice decline implicated several drivers: anomalous meridional heat advection, a deepened ASL, a strongly negative Southern Annular Mode (SAM), and subsurface ocean warming. Remote influences from ENSO, the Indian Ocean Dipole, stratospheric circulation anomalies, and tropical SST variability were also linked. In contrast, the 2022 and 2023 minima occurred with positive SAM but co-occurred with anomalously warm subsurface ocean conditions. A lagged influence of a deepened ASL via enhanced sea-ice export and ice–albedo feedback has been proposed, and the 2023 minimum followed strong positive SAM and La Niña years. These contrasting contexts motivate a detailed assessment of the specific preconditioning and processes that led to the 2023 record low.

Data and models: (1) Observations: NSIDC daily 25-km gridded sea-ice concentration (SIC; NSIDC-0081) from 1979–2022 and near-real-time since January 2023; NSIDC daily sea-ice extent to compare annual cycles. (2) Reanalyses: ERA5 monthly means (1979–2023) for sea-level pressure (SLP), 2-m air temperature, 10-m wind, 1000-hPa temperature and winds, and surface heat flux components (shortwave, longwave, sensible, latent). Temperature advection at 1000 hPa is computed from horizontal winds and temperature gradients. (3) Regional coupled modeling: a Pan-Antarctic Regional Ocean–sea ice model (PAROC) based on NEMO3.6 with LIM3.6 sea ice, covering the Southern Ocean south of 30°S at ~1/4° horizontal resolution (24 km at 30°S to ~14 km near 60°S) with 75 vertical levels. The model is forced by ERA5 atmospheric reanalysis (Jan 1980–Feb 2023) and ORAS5 ocean reanalysis at open boundaries. Model performance is evaluated against NSIDC and GIOMAS, showing high consistency in the Amundsen–Ross Sea during autumn–spring and acceptable summer skill with modest biases. Analyses: Seasonal anomaly composites (MAM, JJA, SON, DJF) of SLP, temperature, advection, and mean seasonal ice concentration (m.s.i.c). Regional focus on Amundsen–Ross Sea (120°W–180°), split into onshore (R1: 70°–80°S) and offshore (R2: 60°–70°S). Sea-ice thickness (SIT) tendency budget terms are diagnosed (growth: basal, lateral, dynamic; melt: surface, lateral, basal, and dynamic/transport) for Oct–Dec 2022 versus climatology (1981–2010). Sea-ice volume (SIV) fluxes out of R1 are computed from modeled SIC, SIT, and ice velocity across defined gates. Surface energy budget diagnostics from ERA5 are decomposed to attribute net surface heat flux anomalies to shortwave components and albedo changes. Spatial patterns of SIC anomalies and retreat rates are used to locate sectors driving SIE changes, particularly the northeastern Ross Sea.

- The 2022–2023 SIE evolution combined two features: (1) persistent, below-average SIE from March 2022 through winter 2022, indicating strong preconditioning; and (2) anomalously rapid retreat in December 2022 (maximum daily SIE loss among compared years), accelerating loss into the February 2023 minimum. - On 21 February 2023, SIE hit 1.79 million km², 6.8% lower than 2022 and ~30% below the 1981–2010 climatology. Negative SIC anomalies were nearly circumpolar at the minimum, with strongest deficits in the Amundsen, Bellingshausen, and Ross seas; the western Indian Ocean was an exception with minor contribution to total SIE. - December 2022 retreat was dominated by strong SIC losses in the northeastern Ross Sea; the SIC reduction within 70°–80°S, 120°W–180° (R1) accounted for about 37% of the total February SIE reduction. - Seasonal drivers: In autumn 2022, a slightly deepened ASL and warm-air advection from lower latitudes produced positive near-surface temperature anomalies in the eastern Pacific sector, inhibiting offshore ice growth and thinning coastal ice in the Ross Sea. In winter, the ASL deepened near the Bellingshausen Sea with warm advection there and cold advection over the Ross Sea. - SIT budget analysis (Oct–Dec 2022): In R1 (onshore), transportation-induced melt dominated total SIT tendency from October to late November (exceeding twice the climatology), evidencing strong sea-ice export. From mid-November, anomalous surface melt tripled climatology, followed by dominant oceanic melt (basal + lateral) until late December as SIV diminished. In R2 (offshore), transportation-induced growth was notably larger than climatology in Oct–Nov, with enhanced surface and slightly elevated oceanic melt in late Nov–mid Dec. - The net surface heat flux anomalies over R1 indicate increased absorption dominated by net shortwave anomalies; decomposition attributes this to reduced surface albedo, consistent with a positive ice–albedo feedback triggered by expanded open water after enhanced export. - ASL behavior: The ASL remained deeper than climatology through winter–summer; in Oct–Nov it was also shifted eastward, bringing strong southerlies over R1 that maximized outward SIV transport, expanded coastal open water, lowered albedo, and warmed the upper ocean. This preconditioning primed extreme summer melt despite a strongly positive SAM. - The December 2022–February 2023 evolution thus reflects synergy between atmospheric circulation (ASL position/depth and advection), dynamic ice export, and thermodynamic feedbacks (albedo–shortwave), with regional contrasts across sectors.

The analysis links the 2023 record-low SIE to a chain of processes operating over the preceding year. A deeper and eastward-shifted ASL during spring enhanced southerly winds along the Amundsen–Ross coast, exporting ice northward and opening coastal polynyas. This reduced surface albedo and increased shortwave absorption, warming the upper ocean and amplifying melt rates as summer approached. The SIT tendency budgets quantify the transition from dynamically driven SIV loss (export) to enhanced surface and then ocean-driven melt, while the offshore region accrued ice from transport before also melting anomalously. The findings reconcile the apparent inconsistency of strong summer melt under a positive SAM by emphasizing the primacy of preconditioning and feedbacks over instantaneous atmospheric forcing. Sectoral differences and comparisons with 2016–2017 and 2022 suggest a non-stationary, mean-state-dependent ASL–sea ice relationship, where the timing, depth, and longitudinal position of the ASL modulate whether export- and albedo-driven feedbacks can be activated to drive rapid retreat.

The summer 2023 Antarctic SIE minimum followed the 2022 record and was characterized by year-round negative anomalies, an exceptional December 2022 retreat focused in the northeastern Ross Sea, and widespread summer deficits. In the Amundsen–Ross sector, a deeper and eastward ASL drove strong southerlies that exported coastal ice northward, produced coastal open water, decreased albedo, and enhanced solar absorption. Subsequent ice–albedo feedback and oceanic melt amplified the initial dynamically forced loss, culminating in the record low. Long-term SIV from GIOMAS and PAROC mirrors SIE behavior, suggesting potential shifts toward lower sea-ice states since 2016, though confirmation requires improved SIT observations. Future work should clarify the evolving ASL–sea ice coupling, quantify the roles of subsurface ocean heat and large-scale modes (SAM, ENSO, IOD, stratospheric variability), and determine causality between thickness and extent changes versus common external forcings.

- Sea-ice thickness observations remain uncertain; satellite-altimetry-derived SIT has sizable errors, limiting direct validation of modeled SIT budgets. - The regional model, while skillful in autumn–spring and capturing key summer features, shows modest summer biases and is not perfect; some diagnostics (e.g., summer SIC/SIT anomalies) have lower, though statistically significant, correlations with observations. - Attribution among intertwined drivers (ASL variability, SAM state, heat advection, subsurface ocean heat) remains challenging; the ASL–sea ice relationship appears non-stationary and mean-state dependent. - The apparent co-variability between sea-ice extent and thickness complicates assessment of causality versus shared external forcing; more accurate, sustained SIT measurements are needed to resolve these links.