Schüller et al. (2025) Quantifying coupling errors in atmosphere-ocean-sea ice models: A study of iterative and non-iterative approaches in the EC-Earth AOSCM

Identification

• Journal: Geoscientific model development
• Year: 2025
• Date: 2025-11-27
• Authors: Valentina Schüller, Florian Lemarié, Philipp Birken, Éric Blayo
• DOI: 10.5194/gmd-18-9167-2025

Research Groups

• Lund University, Lund, Sweden
• Univ. Grenoble Alpes, Inria, CNRS, Grenoble INP, LJK, Grenoble, France

Short Summary

This study quantifies numerical coupling errors in atmosphere-ocean-sea ice models using iterative Schwarz waveform relaxation (SWR) methods, revealing that standard non-iterative coupling introduces substantial errors in atmospheric and sea ice surface temperatures, often due to discontinuous physics parameterizations.

Objective

• Do iterative coupling methods for a given model converge?
• If so, how large is the coupling error of state-of-the-art coupling algorithms?
• If not, which components cause non-convergence?

Study Configuration

• Spatial Scale: Coupled single column model (SCM), one-dimensional in space, simulating physical processes in a vertical column. Atmospheric vertical resolution is approximately 20 meters near the air-sea interface, decreasing towards the top of the atmosphere (80 kilometers). Oceanic vertical resolution is approximately 1 meter near the surface, increasing with depth.
• Temporal Scale: Multi-day simulations (e.g., four-day and two-day experiments). Coupling time step: 60 minutes. Model time steps for ocean, sea ice, and atmosphere: 15 minutes.

Methodology and Data

• Models used: EC-Earth coupled atmosphere–ocean single column model (AOSCM), Open Integrated Forecasting System (OpenIFS) for atmosphere, Nucleus for European Modelling of the Ocean (NEMO) for ocean and sea ice, Louvain-La-Neuve sea Ice Model (LIM3) for NEMO 3.6, Sea Ice modelling Integrated Initiative (SI3) for NEMO 4.0.1. Coupling software: OASIS3-MCT. Iterative coupling algorithm: Schwarz waveform relaxation (SWR).
• Data sources: ERA-Interim (atmospheric forcing for PAPA station), CMEMS Global Ocean Physics Reanalysis (NEMO initialization for PAPA station), ECMWF atmospheric reanalysis (ERA5, atmospheric forcing for MOSAiC expedition), Arctic Ocean Physics Reanalysis (LIM/SI3 initialization for MOSAiC expedition).

Main Results

• In ice-free conditions, the SWR algorithm converges robustly, allowing for quantification of coupling errors. Atmospheric boundary layer temperature coupling errors can reach several degrees Celsius (e.g., 1 °C in a four-day simulation, maximum 3.99 °C in two-day simulations), linked to the discontinuous convective mass flux scheme in OpenIFS. Sea surface temperature (SST) errors are generally small (usually below 0.1 °C, maximum 0.58 °C).
• In ice-covered conditions, the EC-Earth 3 AOSCM (OpenIFS cy40r1–NEMO 3.6) exhibited strong, unphysical oscillations (up to 20 °C for 10 meter atmospheric temperature), indicating issues with the LIM3 sea ice component.
• The latest development version (OpenIFS cy43r3–NEMO 4.0.1 with SI3) shows significantly improved behavior, but non-convergence of SWR still occurs when sea ice approaches melting conditions.
• This non-convergence is attributed to discontinuous jumps in the SI3 ice albedo parameterization at 0 °C. Regularizing this parameterization by smoothing the transition resolves the non-convergence issues.
• With the regularized albedo parameterization, substantial coupling errors are observed for atmospheric temperature (maximum 7.48 °C) and ice surface temperature (maximum 4.66 °C) after two days. SST and atmospheric humidity errors remain small.
• The atmosphere-first coupling algorithm generally performs best for ocean and sea ice variables, but no single non-iterative algorithm consistently minimizes coupling errors across all model components.

Contributions

• First study to extend Schwarz waveform relaxation (SWR) analysis to atmosphere-ocean-sea ice coupling, including the sea ice component.
• Quantification of coupling errors in the EC-Earth AOSCM for both ice-free and ice-covered conditions.
• Identification of specific discontinuous physics parameterizations (convective mass flux in OpenIFS and sea ice albedo in SI3) as primary sources of large coupling errors and SWR non-convergence.
• Development of a reusable Python wrapper for SWR implementation based on the OASIS3-MCT coupler, minimizing modifications to model components.
• Demonstration that smoothing discontinuous parameterizations can significantly improve the robustness and convergence of coupled models.

Funding

• Strategic Research Area “ModElling the Regional and Global Earth system” (MERGE)
• Kungliga Fysiografiska Sällskapet i Lund (grant no. 20240032)
• eSSENCE research program
• French National Research Agency (ANR) under project ANR-23-CE56-0006-01 (MOTIONS)
• Swedish Research Council, Forte, Formas, and Vinnova (for publication)

Citation

@article{Schüller2025Quantifying,
  author = {Schüller, Valentina and Lemarié, Florian and Birken, Philipp and Blayo, Éric},
  title = {Quantifying coupling errors in atmosphere-ocean-sea ice models: A study of iterative and non-iterative approaches in the EC-Earth AOSCM},
  journal = {Geoscientific model development},
  year = {2025},
  doi = {10.5194/gmd-18-9167-2025},
  url = {https://doi.org/10.5194/gmd-18-9167-2025}
}