Trivedi et al. (2025) An assessment of Antarctic sea-ice thickness in CMIP6 simulations with comparison to the satellite-based observations and reanalyses

Identification

• Journal: ˜The œcryosphere
• Year: 2025
• Date: 2025-12-12
• Authors: Shreya Trivedi, Will Hobbs, Marilyn Raphael
• DOI: 10.5194/tc-19-6771-2025

Research Groups

• Department of Geography, University of California, Los Angeles, USA
• Australian Antarctic Program Partnership, Institute for Marine and Antarctic Studies, University of Tasmania, nipaluna/Hobart, Australia

Short Summary

This study assesses the spatio-temporal variations and biases in Antarctic sea-ice thickness (SIT) and volume (SIV) simulated by 39 CMIP6 models against satellite observations and reanalyses, finding that models replicate seasonal cycles but generally underestimate absolute thickness and project seasonally asymmetric declines under future warming.

Objective

• To assess the spatio-temporal variations and biases in Antarctic sea-ice thickness (SIT) and volume (SIV) simulated by 39 CMIP6 coupled climate models, comparing them with satellite-based observations and reanalyses, and to examine their inter-relationships with sea-ice area (SIA) and future projections under a high-emission scenario.

Study Configuration

• Spatial Scale: Circum-Antarctic, Southern Ocean.
• Temporal Scale:
  • Historical simulations and reanalyses: 1979–2014.
  • Satellite observations: 2002–2014.
  • Primary comparison period: 2002–2014.
  • Future projections (SSP5-8.5 scenario): 2015–2100.

Methodology and Data

• Models used: 39 Coupled Climate Models from CMIP6 historical experiments (single ensemble member per model). Specific sea-ice modules like NEMO-LIM3 are noted in some models.
• Data sources:
  • Satellite-derived observations:
    • Envisat-CryoSat-2 (Sea-Ice Climate Change Initiative - SICCI project, 2002–2017, 50 km spatial resolution).
    • National Snow and Ice Data Center (NSIDC) for Sea-Ice Concentration (SIC) (Comiso, 2017) used for satellite-derived SIA.
  • Reanalysis/Synthesis products:
    • Global Ice-Ocean Modeling and Assimilation System (GIOMAS, 1979–2014, 0.8° horizontal resolution).
    • German contribution to the Estimating the Circulation and Climate of the Ocean project Version 3 (GECCO3, 1979–2014, 0.4° nominal horizontal resolution).
  • CMIP6 Model Output: sithick (sea-ice thickness), siconc (sea-ice concentration), areacello (area of individual grid cells over the ocean).
  • Future Scenario: Shared Socio-economic Pathway 5-8.5 (SSP5-8.5).

Main Results

• CMIP6 models generally replicate the mean seasonal cycle and spatial patterns of Antarctic SIT, SIV, and SIA.
• Models show good agreement with satellite observations for SIT maximum in February (summer) but align better with reanalysis products for annual SIT minima (fall).
• Significant negative biases are observed in simulated SIA and SIV annual cycles, with the multi-model mean (MMM) consistently lower than observations/reanalyses.
• Most models simulate thicker sea-ice than observations/reanalyses during summer (January–April) but thinner ice during September (winter).
• The inter-model spread in SIT is larger during summer (November–March) and smaller during April–November.
• CMIP6 models simulate negative trends in Antarctic SIT/SIV, which contrasts with observed positive trends until mid-2015.
• Observations show significant positive trends in SIT/SIV during cooler seasons (winter and spring), which are largely absent in warmer seasons where positive trends are predominantly observed in SIA.
• Future projections under the SSP5-8.5 scenario (2015–2100) indicate pronounced SIT declines primarily confined to cooler seasons, while SIV and SIA exhibit persistent, year-round decreases, with negative anomalies intensifying after approximately 2060.
• Simulated SIA and SIT biases are negatively correlated in February (summer) and positively correlated in September (winter).
• The spatial distribution of SIT biases is closely linked to uncertainties in modeling the ice edge and dynamic processes, such as the Weddell Gyre.
• Models generally underestimate absolute SIT, particularly in deformed-ice regions (approximately 58% less than satellite estimates).
• Certain models exhibit anomalously thick sea-ice (>3 meters) in the western Weddell Sea (along the Antarctic Peninsula) and around the sea-ice edge, potentially due to dynamic processes like ice convergence and misrepresentation of fast-ice.

Contributions

• This study provides the first comprehensive evaluation of Antarctic sea-ice thickness (SIT) and volume (SIV) across 39 CMIP6 models, comparing historical simulations with three distinct sea-ice products (satellite observations and two reanalyses).
• It extends the analysis by examining the inter-relationships between simulated SIT, SIV, and sea-ice area (SIA) during key months (February and September), revealing complex seasonal covariances.
• The research identifies significant seasonal trends in sea-ice parameters and analyzes their evolution, providing insights into how observed post-2016 declines may evolve under future warming scenarios.
• It highlights that SIT variability does not always mirror changes in SIA or volume, underscoring the need for a more nuanced understanding of Antarctic sea-ice responses beyond surface parameters.
• The study emphasizes the critical need for improved representation of Antarctic sea-ice processes (e.g., dynamics, snow-ice formation, ice strength, and atmosphere-ocean-ice coupling) in climate models for more accurate projections of thickness and related volume changes.

Funding

• US National Science Foundation (NSF) under the Office of Polar Programs (grant no. NSF-OPP-1745089).
• Australian Government as part of the Antarctic Science Collaboration Initiative program.
• Australian Research Council Discovery Project 80 (grant no. DP230102994).

Citation

@article{Trivedi2025assessment,
  author = {Trivedi, Shreya and Hobbs, Will and Raphael, Marilyn},
  title = {An assessment of Antarctic sea-ice thickness in CMIP6 simulations with comparison to the satellite-based observations and reanalyses},
  journal = {˜The œcryosphere},
  year = {2025},
  doi = {10.5194/tc-19-6771-2025},
  url = {https://doi.org/10.5194/tc-19-6771-2025}
}