Müller et al. (2025) The ICON-based Earth System Model for climate predictions and projections (ICON XPP v1.0)

Identification

• Journal: Geoscientific model development
• Year: 2025
• Date: 2025-12-03
• Authors: Wolfgang A. Müller, Stephan Lorenz, Trang Van Pham, Andrea Schneidereit, Renate Brokopf, Victor A. Brovkin, Nils Brüggemann, Fatemeh Chegini, Dietmar Dommenget, Kristina Fröhlich, Barbara Früh, Veronika Gayler, Helmuth Haak, Stefan Hagemann, Moritz Hanke, Tatiana Ilyina, Johann Jungclaus, Martin Köhler, Peter Korn, Luis Kornblueh, Clarissa Kroll, Julian Krüger, Karel Castro‐Morales, Ulrike Niemeier, Holger Pohlmann, Iuliia Polkova, Roland Potthast, Thomas Riddick, Manuel Schlund, Tobias Stacke, Roland Wirth, Dakuan Yu, Jochem Marotzke
• DOI: 10.5194/gmd-18-9385-2025

Research Groups

• Max-Planck Institute for Meteorology, Hamburg, Germany
• Deutscher Wetterdienst, Offenbach am Main, Germany
• University of Hamburg, Hamburg, Germany
• ARC Centre of Excellence for Climate Extremes, Monash University, Melbourne, Australia
• Institute of Coastal Systems, Helmholtz-Zentrum Hereon, Geesthacht, Germany
• Deutsches Klimarechenzentrum, Hamburg, Germany
• Institute for Atmospheric and Climate Science, ETH Zürich, Zürich, Switzerland
• Deutsches Zentrum für Luft- und Raumfahrt (DLR), Institut für Physik der Atmosphäre, Oberpfaffenhofen, Germany

Short Summary

This paper introduces ICON XPP (eXtended Predictions and Projections), a new Earth System Model configuration based on the ICON framework, designed for next-generation climate predictions and projections. Two configurations with different spatial resolutions are presented and evaluated, demonstrating performance comparable to CMIP6 models and high computational efficiency for large ensemble simulations.

Objective

• To develop and provide a first evaluation of ICON XPP, a new coupled Earth System Model configuration based on the ICON framework, serving as the baseline for future climate predictions, projections, and contributions to CMIP7.
• To assess the model's ability to represent the basic Earth System state, key modes of climate variability (e.g., ENSO, MJO, QBO), and its overall fidelity compared to existing climate models.

Study Configuration

• Spatial Scale:
  • High-resolution configuration ("80/20"): Atmosphere with ~80 km horizontal grid spacing and 130 vertical levels (model top at 75 km); Ocean with ~20 km resolution and 72 vertical levels.
  • Coarser-resolution configuration ("160/40"): Atmosphere with ~160 km horizontal grid spacing and 90 vertical levels (model top at 75 km); Ocean with ~40 km resolution.
• Temporal Scale:
  • Spin-up and pre-industrial control (CTRL) simulations: 1000 years.
  • Historical (HIST) simulations: 1850–2014 (3 ensemble members).
  • Climate sensitivity experiments (1% CO₂ increase, abrupt 4xCO₂): 150 years.
  • Computational throughput: ~85 simulated years per day (160/40 on 40 nodes), ~45 simulated years per day (80/20 on 100 nodes).
  • Integration time steps: ICON NWP 450 s, ICON O 20 min (80/20) or 30 min (160/40).
  • Coupling interval: 60 min (80/20) or 30 min (160/40) between atmosphere and ocean.

Methodology and Data

• Models used:
  • ICON (ICOsahedral Nonhydrostatic) model framework, including ICON NWP (atmospheric component), ICON O (ocean component), JSBACH (land surface model, versions 3 and 4), HAMOCC6 (ocean biogeochemistry model), and a Hydrological Discharge (HD) model.
  • YAC (Yet Another Coupler) for component coupling.
  • FESIM (Finite-Element Sea Ice Model) for sea-ice dynamics.
• Data sources:
  • CMIP DECK (Diagnostic, Evaluation and Characterization of Klima) experimental framework.
  • CMIP6 historical forcing data.
  • Polar Science Center hydrographic climatology (PHC3.0) for ocean initialization.
  • Observational/reanalysis datasets for evaluation: ERA5 (temperature, precipitation, zonal wind, specific humidity, zonal wind stress), GLODAPv2 (ocean biogeochemistry), GPCP (precipitation), CERES-EBAF (radiation), ESACCI-CLOUD, MODIS, PATMOS-x (cloud properties), ESACCI-WATERVAPOUR (water vapor path), HadCRUT5 (surface temperature), TropFlux (tropical SST, wind stress, heat flux, ENSO), PIOMAS (Arctic sea-ice volume), AVISO (sea surface height).

Main Results

• ICON XPP achieves a stable pre-industrial climate equilibrium with a top-of-atmosphere radiation balance of 0.2 W/m² (160/40) and -0.1 W/m² (80/20), and a global mean near-surface temperature of 13.8–14.0 °C.
• Ocean circulation strengths, such as the Atlantic Meridional Overturning Circulation (AMOC) at 16–18 Sv, are within observed ranges.
• Historical simulations show a global mean temperature increase of approximately 1.2 °C since the 1850s, consistent with reference datasets.
• Regional biases include a pronounced double-ITCZ in the tropical Pacific (up to 6 mm/day wet bias) and a strong warm bias in the Southern Ocean (up to 5 °C higher than observations in 160/40, 3 °C in 80/20).
• Dynamical processes like El Niño/Southern Oscillation (ENSO) exhibit typical characteristics and dynamics, performing within the range of CMIP6 models, though with weaker amplitudes and feedbacks. Tropical waves and the Madden-Julian Oscillation are well captured.
• The 80/20 configuration spontaneously generates a weak Quasi-Biennial Oscillation (QBO) with a period of 17 months, benefiting from higher vertical resolution.
• Northern Hemisphere extra-tropical dynamics show reduced zonal mean zonal wind biases (1–2 m/s for 80/20) and improved representation of jet stream position and transient momentum transports compared to predecessor models.
• Transient Climate Response (TCR) is 1.7 K (160/40) and 1.6 K (80/20), and Equilibrium Climate Sensitivity (ECS) is 2.47 K for both configurations, falling within the lower end of CMIP6 ranges.

Contributions

• Presents the first fully coupled Earth System Model configuration (ICON XPP) based on the ICON framework, integrating operational numerical weather prediction (ICON NWP) components with established climate components for ocean, land, and ocean biogeochemistry.
• Establishes ICON XPP as the foundational model platform for the next generation of climate predictions and projections, including contributions to CMIP7.
• Demonstrates high computational efficiency, enabling long integrations and large ensemble experiments crucial for climate research and operational forecasting.
• Achieves significant improvements in the representation of northern hemisphere extra-tropical dynamics, particularly reduced zonal mean zonal wind biases and more realistic eddy-momentum fluxes, compared to previous ICON-based and MPI-ESM models.
• Provides a comprehensive evaluation of key climate phenomena, including a detailed assessment of ENSO dynamics and its relation to mean state biases, showing performance comparable to CMIP6 models.

Funding

• German Ministry of Education and Research (BMBF):
  • Coming Decade (grant no. 01LP2327E)
  • CAP7 (grant no. 01LP2401C)
• Hans Ertel Centre for Weather Research (HErZ):
  • OceanWeather (grant no. 4823DWDP1B)
• Deutscher Wetterdienst (DWD) "Innovation Program for Applied Research and Development – IAFE":
  • ICON-Seamless VH 4.7 project
  • POINTS project
• German Research Foundation (DFG) under the Excellence Strategy – EXC 2037 "CLICCS – Climate, Climatic Change, and Society" (no. 390683824)
• DFG Research Unit VolImpact (FOR2820, no. 398006378)
• European Space Agency (ESA) project STATISTICS
• Australian Research Council (ARC) Centre of Excellence for Climate Extremes (grant no. CE170100023)
• ETH Postdoctoral Fellowship program
• Swiss National Supercomputing Center (CSCS) under project ID s1283
• European Union's Horizon Europe research and innovation program:
  • ESM2025 (grant no. 101003536)
  • ASPECT (grant no. 101081460)
  • Q-Arctic (grant no. 951288)
• Max Planck Society (for article processing charges)

Citation

@article{Müller2025ICONbased,
  author = {Müller, Wolfgang A. and Lorenz, Stephan and Pham, Trang Van and Schneidereit, Andrea and Brokopf, Renate and Brovkin, Victor and Brüggemann, Nils and Chegini, Fatemeh and Dommenget, Dietmar and Fröhlich, Kristina and Früh, Barbara and Gayler, Veronika and Haak, Helmuth and Hagemann, Stefan and Hanke, Moritz and Ilyina, Tatiana and Jungclaus, Johann and Köhler, Martin and Korn, Peter and Kornblueh, Luis and Kroll, Clarissa and Krüger, Julian and Castro‐Morales, Karel and Niemeier, Ulrike and Pohlmann, Holger and Polkova, Iuliia and Potthast, Roland and Riddick, Thomas and Schlund, Manuel and Stacke, Tobias and Wirth, Roland and Yu, Dakuan and Marotzke, Jochem},
  title = {The ICON-based Earth System Model for climate predictions and projections (ICON XPP v1.0)},
  journal = {Geoscientific model development},
  year = {2025},
  doi = {10.5194/gmd-18-9385-2025},
  url = {https://doi.org/10.5194/gmd-18-9385-2025}
}