IFS model description

Both IFS-FESOM and IFS-NEMO coupled models use as their atmospheric component an updated version of the Integrated Forecasting System (IFS) Cycle 48R1 with modifications from nextGEMS and the Climate Digital Twin [Rackow et al., 2025] [Segura et al., 2025]. The atmospheric component is coupled to the ecWAM wave model (ecWAM repository), which runs alongside IFS at 0.125° resolution, and includes the ECLand land-surface scheme [Boussetta et al., 2021] [Rackow et al., 2025], ensuring a consistent representation of atmosphere–land–wave interactions.

Atmosphere component

The atmospheric component of IFS employs a hydrostatic, semi-Lagrangian, semi-implicit dynamical core. The horizontal discretization uses a triangular cubic-octahedral (TCo) grid, with TCo2559 (~4.4–4.5 km) for control, historical, and SSP3-7.0 simulations, and TCo1279 (~9 km) for storyline and other science-driven simulations. The vertical discretization includes 137 levels, covering the full troposphere and stratosphere.

Key physical parameterizations include a mass-flux convection scheme (with strongly reduced deep convection at 4.4 km to improve tropical convective organization and the ITCZ), prognostic cloud microphysics (liquid water, ice, rain, snow, and cloud fraction), an eddy-diffusivity mass-flux scheme for turbulence and boundary-layer processes, and gravity-wave drag from both orographic and non-orographic sources.

Key modifications introduced on top of 48R1 include the activation of a tracer global mass fixer for all prognostic hydrometeors — cloud liquid, ice, rain and snow — to mitigate water and energy non-conservation, as well as cloud and microphysics adjustments within observational uncertainty ranges to better reproduce the observed top-of-atmosphere radiation balance, and the development of a reduced cloud base mass flux approach to improve cloud organisation and precipitation characteristics.

Land Component

ECLand includes soil, lakes, snow, and vegetation based on the ESA-CCI land cover dataset, with an urban scheme driven by the ECOCLIMAP-SG global urban map [McNorton et al., 2021] [McNorton and Di Giuseppe, 2023]. Advanced snow and soil physics account for multi-layer snow, permafrost processes, and revised river discharge, ensuring a consistent treatment of land–atmosphere interactions.

ecWAM runs on a reduced lat–long 0.125° grid with 36 directions and 36 frequencies, following a frequency discretization that allows ocean waves with periods between 1 and 28 s to be represented.

The main updates with respect to cycle 48R1 were an upgrade of the land surface scheme to ECLand v1.1.0 as in cycle 49R1 [ECMWF, 2024], incorporating kilometre-scale land use and land cover information based on ESA-CCI products at up to 300 m resolution, improved post-processing of 2 m temperature to reduce warm biases under stable conditions, and an urban scheme representing the distinct influence of urban areas on the boundary layer and surface energy balance [Boussetta et al., 2021] [ECMWF, 2024].

Initial conditions and forcing

The initial conditions for the state of land and atmosphere come from ECMWF’s operational analysis of the 20th of January 2020, which provides the best estimate, in near-real time, of the state of the atmosphere for that given date. This operational analysis is the result of a blend of in-situ and remote sensing observations available in near-real time with IFS simulations via the 4D-Var data assimilation system [ECMWF, 2023].

A time-varying aerosol climatology from the European project CONFESS is used to represent tropospheric aerosol species forcings over the historical 1990–2014 period. Beyond 2015, a linear scaling of CONFESS aerosol species is applied derived from the evolution of MACv2-SP [Stevens et al., 2017] forcings under the SSP3-7.0 scenario. This approach ensures scenario consistency while leveraging the strengths of the CONFESS dataset to improve the representation of future aerosol impacts.

The experiments additionally use greenhouse gas concentrations (which include CO2, CH4 and N2O), solar irradiance and ozone historical and SSP3-7.0 forcings produced for CMIP6.