Isotopic evidence for an intensified hydrological cycle in the Indian sector of the Southern Ocean

The study investigates whether and how the hydrological (atmospheric water) cycle has intensified in the Southern Ocean under recent warming, focusing on the Indian sector (40°E–90°E). Observational evidence is challenging due to sparse measurements and the overlapping influences of precipitation–evaporation (P−E), sea ice processes, and glacial meltwater. The authors aim to disentangle these contributions using concurrent surface salinity and oxygen isotope (δ18O) observations spanning 1993–2021, restricted to austral summer. They hypothesize that intensified atmospheric freshwater fluxes (notably increased net precipitation at high latitudes and decreased net freshwater into the subtropics) and changing cryospheric freshwater fluxes (sea ice and glacial melt) have produced contrasting salinity and δ18O trends across the Subantarctic Front. Establishing robust observational constraints on these processes is important for understanding Southern Ocean stratification, circulation, sea-ice regime, and global climate feedbacks.

Prior work has documented a complex pattern of Southern Ocean change: surface cooling south of the ACC alongside marked warming to the north extending to depth; widespread freshening in subpolar surface waters and at intermediate depths. These changes relate to the ocean’s role in heat and carbon uptake and to stratification effects that influence circulation and sea ice. Several studies attribute subpolar freshening to an amplification of the hydrological cycle with increased net precipitation at high latitudes. Alternative or additional mechanisms include increased freshwater from ice shelf/iceberg melt and intensified sea-ice freshwater transport. Climate models attribute part of recent salinity changes to anthropogenic forcing but often poorly represent Antarctic ice-shelf melt variability, Antarctic sea ice cover/trends, and net precipitation biases. Therefore, observational constraints that separate meteoric (precipitation/glacial) from sea ice freshwater sources are needed; δ18O is a useful tracer because meteoric inputs are δ18O-depleted whereas sea ice formation/melt has minimal isotopic fractionation and primarily affects salinity.

Data and study region: Surface salinity (absolute salinity, TEOS-10) and δ18O observations in the Indian sector of the Southern Ocean (40°E–90°E, 30°S–60°S) during austral summers (Dec–Feb) from 1993 to 2021 were compiled from the OISO program (1998–2021) and additional hydrographic sections (1993–1998). Samples were taken underway at ~5–7 m and from stations within 0–50 m. Analytical methods and calibration: δ18O was measured using IRMS (Isoprime with Multiprep; pre-2010) and CRDS (Picarro L2130-i; mainly post-2010). CRDS analyses used a stainless-steel liner to mitigate salt effects; a post-processing “sea salt effect” correction was applied. Cross-dataset calibration corrected nonphysical offsets by enforcing stability of deep water δ18O (neutral density 28.15–28.3 kg m−3) across datasets. The δ18O uncertainty is ~0.06‰. Salinity was measured via salinometer, calibrated CTD, and thermosalinograph data. Coordinate framework: To account for the meandering ACC and dynamics, the analysis uses a streamwise coordinate system based on mean dynamic topography (AVISO). Observations were binned on an irregular MDT grid (0.1–0.3 m resolution; finer within the ACC). Climatological means, anomalies, and trends were computed in this streamwise framework and mapped back to geographic space as a pseudo-latitude (lat) for readability. Meridional structure and anomaly correlation: Zonal-mean meridional profiles of surface salinity and δ18O were constructed. North of the SAF-N (lat ~46°S), salinity and δ18O are highly correlated and align along a linear mixing line (slope ~0.5 ± 0.01 ‰ kg g−1), indicating dominant control by E−P processes. South of 46°S, salinity–δ18O correlation weakens, consistent with additional influences from sea ice and meteoric inputs. Trend estimation: For broad sectors (north and south of 46°S), annual medians of anomalies were computed and a weighted linear regression was fit to annual medians (weights w = 1/IQR²). Uncertainties were derived via bootstrap (50,000 resamples with 80% subsampling). For meridional sections, linear trends and standard errors were computed in each streamwise bin; robustness is indicated relative to standard error thresholds. Freshwater budget framework: Long-term changes in mixed-layer salinity and δ18O were related to freshwater flux components via budgets integrated over the seasonal cycle. The salinity budget yields hΔS = (SFWF − So)ΔFWF, and similarly for δ18O. Decomposing ΔFWF into P−E (FP−E), sea ice (FS), and glacial meltwater (FIS) gives two equations (for salinity and δ18O) that are solved for ΔFP−E using best estimates (and uncertainties) for sea-ice and glacial meltwater changes and endmember properties (Tables 1–2). Horizontal advection changes are neglected due to the streamwise approach and the lack of large meridional front shifts over recent decades. To propagate uncertainties and potential approximations (including small frontal shifts), a Monte Carlo procedure repeated 50,000 times varied all inputs within their uncertainties and retained solutions where ΔFP−E estimates from salinity and δ18O budgets agreed within ±5 mm yr−1 per decade. External constraints for freshwater components: Sea-ice freshwater flux trends for 1993–2008 were taken from published estimates combining sea-ice concentration, drift, and thickness via a mass-balance approach; over 60°S–46°S (subpolar band) the mean trend is −12 ± 4 mm yr−1 per decade (salinity-increasing). Glacial freshwater changes were constrained from IPCC-assessed Antarctic mass loss (grounding-line discharge changes) and ice-shelf thinning/ice-front retreat, giving a plausible total increase equivalent to ~150–550 Gt yr−1 per decade (1993–2021), translating to ~5–17 mm yr−1 per decade when spread over the subpolar mixed-layer area; this is considered an upper-bound for open-ocean influence. Endmember salinity and δ18O ranges for mixed layer, sea ice, glacial meltwater, and P−E were set from literature. The northern subtropical sector was analyzed primarily in terms of Δ(P−E), given minimal sea-ice/glacial influence at the surface there.

- Strong, contrasting trends across the Subantarctic Front (~46°S): - Subtropics (north of 46°S): surface salinity increased by 0.06 ± 0.07 g kg−1 per decade and δ18O increased by 0.03 ± 0.04‰ per decade (1993–2021). Salinity and δ18O anomalies are tightly correlated (r > 0.8), consistent with dominant control by E−P along a linear mixing line (slope ~0.5 ± 0.01 ‰ kg g−1). - Subpolar (south of 46°S): surface salinity decreased by −0.02 ± 0.01 g kg−1 per decade and δ18O decreased by −0.01 ± 0.02‰ per decade. Salinity–δ18O anomalies show weak correlation, indicating additional influences from sea ice and meteoric inputs. - Attribution of freshwater forcing changes: - Subpolar sector: Freshening is largely driven by an increase in net precipitation (approximately doubled), while a decline in sea-ice melt (leading to a salinifying tendency) is largely offset by increased glacial meltwater at these latitudes. Quantitatively, sea-ice freshwater flux trend is −12 ± 2 mm yr−1 per decade (salinity-increasing), glacial meltwater contributes +11 ± 4 mm yr−1 per decade; the net P−E change inferred from the dual-budget framework imparts a dominant freshening, equivalent to a salinity tendency of about −0.02 ± 0.008 g kg−1 per decade. - Subtropics: Changes are explained primarily by Δ(P−E), with the dual-budget solution indicating an increase in (E−P) or decrease in net freshwater input; inferred Δ(P−E) is ~+33 ± 11 mm yr−1 per decade (sign convention positive into ocean), consistent with observed salinification and δ18O increase. - Limited direct open-ocean impact of enhanced Antarctic glacial meltwater over 1993–2021: estimated contribution to subpolar open-ocean salinity trend is small (≈ −0.008 ± 0.003 g kg−1 per decade), consistent with coastal confinement and mixing that attenuates δ18O-depleted signals offshore. - Sea-ice decline in the northern sea-ice sector/subpolar Indian Ocean produces a salinity increase (~+0.008 ± 0.002 g kg−1 per decade), but this is outweighed by the stronger freshening from increased P−E, yielding net subpolar freshening. - Independent large datasets of mixed-layer properties (1970–2018) show similar meridional salinity-change patterns and magnitudes, supporting the robustness of the trends and the streamwise-coordinate approach.

The findings directly address the research question by providing observational evidence that the atmospheric water cycle has intensified in the Indian sector of the Southern Ocean, with opposing salinity and δ18O trends across the Subantarctic Front. The strong co-variation of salinity and δ18O north of 46°S and the derived P−E changes indicate reduced net freshwater input (or enhanced evaporation) in the subtropics, consistent with a strengthened subtropical dry regime. In contrast, the subpolar freshening and δ18O decrease, together with weak salinity–δ18O correlation, reveal combined influences of increased net precipitation and cryospheric freshwater flux changes. However, the glacial meltwater contribution to open-ocean properties is small compared to P−E, and the salinifying effect of sea-ice decline is more than compensated by enhanced meteoric input, yielding net freshening. These results support the broader picture of a globally intensifying hydrological cycle and highlight the sensitivity of Southern Ocean stratification and circulation to changes in surface freshwater forcing. The approach using concurrent δ18O and salinity proves powerful for disentangling the roles of meteoric versus sea-ice sources, offering observational constraints for improving climate model representations of polar processes and their attribution of anthropogenic influence.

This study provides quantitative observational evidence for an intensified hydrological cycle in the Indian sector of the Southern Ocean since the early 1990s, characterized by subtropical salinification and subpolar freshening with consistent δ18O trends. Using a streamwise-coordinate framework and dual salinity–δ18O budgets, the authors attribute subpolar freshening primarily to increased net precipitation, with limited direct open-ocean influence from enhanced Antarctic glacial meltwater and a compensating salinifying effect from reduced sea-ice melt. In the subtropics, changes are explained by P−E alone. The work demonstrates the value of long-term, concurrent δ18O and salinity observations (e.g., OISO) to disentangle complex freshwater processes and provides targets for model development and attribution studies. Future research should extend this combined tracer approach to other Southern Ocean sectors to resolve spatial heterogeneity, continue sustained observations to detect emerging glacial meltwater signals offshore, and integrate improved cryospheric freshwater flux reconstructions to refine attribution.

- Seasonal and spatial sampling: Observations are limited to austral summer and to the Indian sector (40°E–90°E), potentially missing seasonal and regional variability. - Data sparsity and method assumptions: The Southern Ocean is data sparse; sea-ice freshwater flux constraints used here cover 1993–2008 and are regionally averaged. The freshwater budget neglects changes in horizontal advection; while streamwise coordinates minimize bias from frontal meandering, small front shifts and other approximations remain. - Endmember and flux uncertainties: Endmember salinity and δ18O values for P−E, sea ice, and glacial meltwater, and the magnitudes of sea-ice/glacial flux changes carry uncertainties; a Monte Carlo approach addresses these but residual structural uncertainties persist. - Open-ocean focus: The analysis emphasizes open-ocean responses; coastal processes and shelf exchanges, where glacial meltwater signals are stronger, may not be fully captured offshore. - Reliance on external products: P−E trends from atmospheric reanalyses are inconsistent; although the study infers P−E changes from ocean properties, some external datasets (e.g., MDT, sea-ice fluxes) introduce additional uncertainty.