Maurer et al. (2026) Evaluation of coupled and uncoupled ocean–ice–atmosphere simulations using icon-2024.07 and NEMOv4.2.0 for the EURO-CORDEX domain

Identification

• Journal: Geoscientific model development
• Year: 2026
• Date: 2026-01-15
• Authors: Vera Maurer, Wibke Düsterhöft-Wriggers, Rebekka Beddig, Janna Meyer, Claudia Hinrichs, Ha Thi Minh Ho-Hagemann, Joanna Staneva, Birte-Marie Ehlers, Frank Janssen
• DOI: 10.5194/gmd-19-543-2026

Research Groups

• Deutscher Wetterdienst, Offenbach, Germany
• Bundesamt für Seeschifffahrt und Hydrographie, Hamburg/Rostock, Germany
• Institute of Coastal Research, Helmholtz-Zentrum Hereon, Geesthacht, Germany

Short Summary

This paper evaluates ROAM-NBS, a newly developed regional coupled ocean-ice-atmosphere model (ICON-CLM and NEMOv4.2.0), for the EURO-CORDEX domain, comparing its reanalysis-driven simulations (1979-2021) against uncoupled components and observations. The study finds that ROAM-NBS adequately represents the mean climate, variability, and extreme events, making it suitable for future regional climate projections for Europe, particularly the North and Baltic Seas.

Objective

• To analyze the general performance of the newly developed regional coupled ocean-ice-atmosphere model (ROAM-NBS) for the EURO-CORDEX domain, assessing if coupling introduces significant degradation or drift in mean climate and how well it represents key climate-relevant features (e.g., sea surface temperatures, near-surface air temperature, surface heat fluxes, stratification, extreme sea level) and extreme events compared to uncoupled components and observations.

Study Configuration

• Spatial Scale: EURO-CORDEX domain for the atmosphere (approximately 12.1 km horizontal resolution); North and Baltic Sea (NBS) for the ocean (2 nautical miles, approximately 3.7 km, horizontal resolution).
• Temporal Scale: Reanalysis-driven simulations for 1979–2021 (evaluation period 1979–2020). Ocean component spin-up from 1974–1978.

Methodology and Data

• Models used:
  • ROAM-NBS (Regional Ocean-Atmosphere Model - North and Baltic Sea)
  • Atmosphere: ICON-CLM (ICON atmosphere model in Climate Limited-area Mode) version icon-2024.07
  • Ocean: NEMOv4.2.0 (Nucleus for European Modelling of the Ocean)
  • Sea Ice: SI3 (Sea Ice modelling Integrated Initiative) thermodynamic sea ice model
  • Coupler: OASIS3-MCT (flux-based approach)
• Data sources:
  • Reanalysis for forcing: ERA5 (atmosphere), ORAS5 (ocean lateral boundary conditions).
  • Observations for evaluation:
    • Sea Surface Temperature (SST): Copernicus satellite observations (CMEMS).
    • Sea Ice Concentration: Copernicus (CMEMS).
    • Surface Salinity: Copernicus in-situ and satellite observations (CMEMS).
    • Ocean Temperature/Salinity Profiles: Copernicus in-situ profile data (CMEMS), Baltic thalweg level 4 dataset.
    • Near-surface Air Temperature (tas, tasmin, tasmax): E-OBS dataset (over land).
    • Surface Radiation: CERES satellite data.
    • Precipitation: E-OBS, GPM (IMERG Final Run v07).
    • Wind Speed: DWD station observations (German coast), FINO1 platform measurements (100 m height).
    • Sea Surface Height (SSH) / Storm Surge: GESLAv3.0 observational data.
    • Marine Heatwaves (MHWs): In-situ observations, Copernicus reanalysis data.
  • Other input data: River runoff (German Bundesanstalt für Gewässerkunde), tidal constituents (FES2014), CMIP6 ozone dataset, ECHAM6 aerosol climatologies, satellite-derived cloud droplet number concentrations, Harmonized World Soil Database v2.0, EMODnet bathymetry.

Main Results

• All model setups (ROAM-NBS, ICON-CLM, NEMO-NBS) generally represent the mean climate well.
• The area-averaged Sea Surface Temperature (SST) bias is approximately ±0.5 K, with regional variations (e.g., cold bias up to 2 K in shallower European North West Shelf in colder seasons, positive bias in summer, negative in Baltic Sea). ROAM-NBS is slightly warmer than NEMO-NBS in summer (0.3 to 0.4 K). No accumulation of SST errors over the simulation period.
• Over land, near-surface air temperature shows a negative bias for diurnal mean and maximum temperatures, with a small positive bias for diurnal minimum temperatures. Differences between ROAM-NBS and ICON-CLM over land are very small.
• Differences in precipitation and turbulent heat fluxes over the ocean between coupled and uncoupled simulations are directly related to SST differences, but their mean influence on land areas is negligible.
• Wind speed over the ocean is generally underestimated by both ROAM-NBS and ICON-CLM, with ROAM-NBS showing a slightly more pronounced underestimation due to a negative SST bias along the German coast.
• Both ROAM-NBS and NEMO-NBS overestimate sea ice concentration and extent, particularly in the Gulf of Bothnia during spring, linked to cold SST and underestimated salinity.
• Mean temperature and salinity profiles in the Baltic Sea are generally reproduced, but both simulations exhibit cold and fresh biases in deeper layers (e.g., salinity bias up to -4.64 psu in Gotland Deep). Surface salinity is well represented in open waters but underestimated at some coasts.
• Sea surface height and storm surge show high correlation with observations. ROAM-NBS captures extreme SSH maxima more effectively than NEMO-NBS, suggesting improved wind forcing during storms due to coupling. Major Baltic inflow events are qualitatively reproduced but quantitatively underestimated.
• Both simulations capture the inter-annual variability of marine heatwave characteristics reasonably well, with NEMO-NBS showing slightly better alignment with observations in terms of event frequency, maximum intensity, and duration.

Contributions

• Introduces and evaluates ROAM-NBS, one of the first regional climate modeling systems to couple the ICON atmosphere model (icon-2024.07) with the NEMOv4.2.0 ocean model and SI3 sea ice model for the North and Baltic Sea within the EURO-CORDEX domain, utilizing a flux-based OASIS3-MCT coupling approach.
• Represents a methodological advancement over previous regional coupled models by incorporating a later ICON release, a more recent NEMO version with significant updates, higher-resolution EMODnet-based coastal bathymetry, and an enhanced radiation scheme (three-band RGB with prescribed chlorophyll).
• Provides a comprehensive evaluation of the coupled system's performance against its uncoupled components and extensive observational datasets, demonstrating its consistency and robustness in representing mean climate, variability, and extreme events.
• Establishes ROAM-NBS as a suitable tool for generating coupled regional climate projections for Europe, contributing to the EURO-CORDEX ensemble and supporting climate adaptation strategies for German national waters.
• Highlights the advantage of regional coupled models in providing consistent atmosphere-ocean climate information and independence from global climate model projections for regional downscaling.

Funding

• Project CoastalFutures-2 (Germany's Federal Ministry of Research, Technology and Space - BMFTR).
• Project IS-ENES (European Union's Horizon 2020 research and innovation programme, grant agreement no. 824084).
• Deutsches Klimarechenzentrum (DKRZ) project ID bb1338.
• UDAG project (Updating the data basis for adaptation to climate change in Germany).

Citation

@article{Maurer2026Evaluation,
  author = {Maurer, Vera and Düsterhöft-Wriggers, Wibke and Beddig, Rebekka and Meyer, Janna and Hinrichs, Claudia and Ho-Hagemann, Ha Thi Minh and Staneva, Joanna and Ehlers, Birte-Marie and Janssen, Frank},
  title = {Evaluation of coupled and uncoupled ocean–ice–atmosphere simulations using icon-2024.07 and NEMOv4.2.0 for the EURO-CORDEX domain},
  journal = {Geoscientific model development},
  year = {2026},
  doi = {10.5194/gmd-19-543-2026},
  url = {https://doi.org/10.5194/gmd-19-543-2026}
}