EERIE: Ocean Eddy-rich Kilometer-scale Climate Simulation with Integrated Forecasting System (IFS) - Finite volumE Sea Ice-Ocean Model (FESOM2.5): SSP2-4.5 scenario simulation (Version 1)

Project: European Eddy RIch Earth System Models - This project, running from January 2023 to December 2026, will reveal and quantify the role of ocean mesoscale processes in shaping the climate trajectory over seasonal to centennial time scales. To this end EERIE will develop a new generation of Earth System Models (ESMs) that are capable of explicitly representing a crucially important, yet unexplored regime of the Earth system – the ocean mesoscale. Leveraging the latest advances in science and technology, EERIE will substantially improve the ability of such ESMs to faithfully represent the centennial-scale evolution of the global climate, especially its variability, extremes and how tipping points may unfold under the influence of the ocean mesoscale.

Model improvements include new dynamical cores, new components (particularly for sea ice), scale-aware parametrisations and the complementary use of Machine Learning (ML). The technological challenge associated with this ambition is very high. EERIE’s goal is to achieve a simulation speed of up to 5 simulated years per day and to make efficient use (reduction in power consumption by 50%) of the pre-exascale supercomputers now available in Europe. The technological solutions that are to be leveraged in EERIE are the use of reduced numerical precision, Graphic Processing Units (GPUs), ML and reduced Input/Output (I/O). Alongside model improvements, EERIE will develop innovative experimental simulation protocols that are suitable for the mesoscale, to be pioneered on behalf of the global climate modelling community, in preparation for the next Intergovernmental Panel on Climate Change (IPCC) cycle.

(Ref: https://eerie-project.eu/about/)
Summary: The EU project European Eddy RIch Earth System Models (EERIE) aims to advance kilometer-scale Earth System Models (ESMs) to reduce biases associated with low-resolution climate simulations. Its goal is to develop centennial-scale ESMs that explicitly resolve ocean mesoscale processes, thereby improving the representation of long-term climate evolution, variability, extremes, and potential tipping points.
One of these models, IFS-FESOM2-SR, couples the ECMWFs Integrated Forecast System (IFS) atmosphere (9 km resolution) with the FESOM2.5 ocean model (minimum 5 km resolution). The FESOM2.5 ocean model employs an NG5 unstructured triangular grid with 70 depth levels, achieving ~5 km resolution in eddy-rich mid- and high-latitudes and ~13 km in the tropics (Rackow et al., 2025). Its sea-ice component is FESIM (Danilov et al., 2015). The atmospheric model, IFS cycle 48r1 from ECMWF, uses a Tco1279 (~10 km) octahedral grid with 137 vertical levels. The setup follows Rackow et al. (2025) except for deep convection, where the operational IFS scheme is used instead of the modified reduced cloud-base mass flux version.
Following the HighResMIP protocol (Haarsma et al., 2016), the main simulations were preceded by a 50-year spin-up period using 1950 CMIP6 forcing. From the spin-up’s final state, two simulations were launched in parallel: a control run and a historical run using CMIP6 forcings. According to the HighResMIP protocol, the control simulation aimed to assess any potential drift within the simulation, enabling us to exclude the influence of such drift in order to better understand the impact of changes in radiative forcing over time.
After completion of the historical simulation, the experiment was extended along the SSP2-4.5 scenario pathway until 2050 to estimate near-future climate change using a long-term climate simulation at the kilometer scale. For this purpose, CMIP6 scenario forcings were used. Tropospheric aerosol estimates are based on the MACv2 aerosol forcing, which was subsequently adjusted to ensure compatibility with the CONFESS aerosol forcing used during the historical simulation.