In 2025, IMPRESSION-ESM made a significant contribution to the development of the new CNRM and IPSL climate model versions to be used for the IPCC Assessment Fast Track (AFT, simulations planned in 2026) and the CMIP7-Science exercise (tentatively in 2028). These contribution span both individual components and the assembly of the full physical modeling system. The project also supports longer term model developments, as discussed below.

WP1 - Atmospheric processes

An ARPEGE-Climat version consistent with its weather prediction counterpart and calibrated towards climate metrics was finalized and validated. The implementation of a new shallow convection scheme is ongoing, as well as the revisit of cloud-radiation interactions. The IPSL climate model AFT version now uses the new atmospheric dynamical core DYNAMICO. Intensive parameterization developments in LMDZ focused on cold clouds, boundary-layer mixed-phase clouds, precipitation, cold pools and shallow-to-deep convection processes and resulted into 5 PhD thesis defense in fall 2025s. The tuning of an updated LMDZ version that includes these recent developments is still in progress. The evaluation of the shallow-to-deep convection transition in ARPEGE and LMDZ revealed contrasting behaviors and a transition that is too rapid in both models (A. Jardot’s internship). The DEPHY single-column modeling tools were further consolidated and validated (article to be submitted in 2026). N. Nanou (LMD) and L. Maheux (CNRM) started their IMPRESSION-ESM-funded PhD on the coupling between convection and subgrid-scale water distribution and the parameterization of deep convective clouds, respectively.

WP2 - Land-atmosphere interactions and land processes

The four cases designed to evaluate land-atmosphere (L-A) interactions under homogeneous land conditions in single-column models and large-eddy simulations (LES) were published (Bernard et al., 2025). The implementation of the simplified surface model (SSM, A. Maison) in LMDZ and Meso-NH-SURFEX now provides a suitable modeling framework to evaluate the L-A coupling at the process level. We are moving towards heterogeneous surfaces: design of a reference case using the MOSAI forest-to-maize field transition (G. Pain’s internship); land surface model mosaic extension to the atmosphere (E. Bernard); use of more idealized frameworks with the SSM (Maison et al.).

ORCHIDEE developments towards AFT and CMIP7 versions include i) updates of the roughness and resistance schemes; ii) updates of the canopy radiative transfer; iii) inclusion of permafrost processes (Gaillard et al., 2025); iv) implementation of the ORCHIDEE-ice new snow scheme; v) improved tuning of the surface albedo. Work on the multi-tiling framework focuses on the multi-energy budget. The irrigation parametrization and its links with river routing and vegetation processes was consolidated (Arboleda-Obando et al., 2025). Representing irrigation over Spain improves river flows, though agricultural practices need to be better accounted for, and impacts the L-A coupling at regional/seasonal scales (Tiengou et al., 2025).

WP2 - Snow processes

S. Krishnakumar (LSCE) is currently developing a spectral snow albedo formulation, including light-absorbing particle deposition. A parameterization of blowing snow processes was implemented in LMDZ (Vignon et al. 2026). Conesa et al. (2026) developed an initialization procedure for snow over ice sheets as well as parameterizations of dry-snow compaction and fresh snow density. C. Amory submitted an ANR JCJC proposal (HySIS) targeting the representation of melt–albedo feedbacks and surface hydrology over ice sheets in ORCHIDEE.

A joint IGE-LSCE field visit at Col du Lautaret was organized to allow IMPRESSION-ESM modelers to better understand in situ snow measurement techniques and how to properly use them to evaluate spectral albedo (S. Krishnakumar) and meltwater percolation (PhD G. Pitiot, IGE) parameterization developments. Work is ongoing to include other sites (Lautaret, Svalbard) and controlled cold-laboratory experiments. IMPRESSION-ESM supported the organization of a snow workshop in November 2025 (see report), which stimulated collaborative efforts toward defining common evaluation metrics.

WP3 - Ocean processes

Our efforts focused on publishing an improved parameterization of mesoscale eddy effects and documenting its impacts in global ocean simulations using NEMO3.6 (Torres et al. 2025). The parameterization is the first of its kind to (i) be both energetically and observationally constrained, with parameterized sub-grid eddy kinetic energy in excellent agreement with observations and (ii) represent re-stratification and tracer diffusion effects in a consistent framework. The parameterization code was ported to NEMO versions 4.2 and 5.0 and is now included in the NEMO main development branch, ensuring its dissemination within the two French climate models, and beyond. The parameterization is currently being tested in the CNRM climate model, focusing on its impacts on the Atlantic Meridional Overturning Circulation (AMOC). The Giordani et al. (2020) mass-flux parameterization of oceanic convection is also being tested in the CNRM climate model.

WP3 - Sea-ice processes

The BBM rheology developed at IGE (Brodeau et al. 2024) was implemented in NEMO (C. Rousset, LOCEAN) and is currently being tested, especially through the PhD projects of E. Ortega (LOCEAN) and A. Lambotte (UCLouvain), which compare model results with observations of Antarctic sea ice leads and Arctic landfast ice, respectively. The velocity-restoring approach of Pirlet et al. (2025) is shown to improve the realism of Antarctic landfast ice. Pirlet et al. further show improved representation of water mass transformation in polynya regions. As the velocity restoring approach is not suitable for future climate applications, ongoing developments focus on an interactive iceberg–sea ice force formulation. A workshop on iceberg–sea ice interactions organized in June 2025 and supported by IMPRESSION-ESM led to a synthesis paper arguing the need to represent these interactions in climate models (Vancoppenolle et al.). Vertically resolved salt dynamics in sea ice was implemented and evaluated in NEMO v5 (Ortega et al.), in particular owing to a new compilation of ice core data. Progress is also being made on the sea-ice penalization to represent its embedment into the ocean (Billy et al.), and on snow processes (Brivoal et al.).

WP4 - Ocean-atmosphere interactions

Dehondt et al. investigates the sensitivity of the AMOC to the parameterizations of air–sea turbulent fluxes within the IPSL climate model, highlighting the critical role of the surface roughness length at high wind speeds. Compared to the direct effect of the wind stress parameterization, local and remote feedbacks dominate the sensitivity of surface fluxes and ocean dynamics in the North Atlantic. The air–sea flux formulation affects the relative strength of the subtropical and subpolar gyres, thereby controlling heat and salt transport in regions of deep-water formation. Marti et al. (2021) demonstrated that, while a Schwarz iterative algorithm improves the temporal coupling at the ocean-atmosphere interface, it is not computationally efficient. In collaboration with the LJK, Grenoble, a neural network was developed to compute an improved initial guess and thereby accelerate convergence. It is currently being implemented in the IPSL climate model.

WP4 – Sea ice-ocean-iceberg interactions

Olivé Abelló et al. (2025) implemented in the NEMO that icebergs shall inherit the thickness of the ice shelf front from which they calve, resulting to thicknesses far greater than the 250-m maximum previously assumed. Due to their grounding to by shallow underwater ridges, icebergs spend more time on the Antarctic continental shelf, thereby increasing the associated meltwater discharge onto the continental shelf and leading to the acceleration of the ice-shelf melting. Grounded icebergs also anchor the drifting sea ice, acting as a barrier that accumulates thick sea ice to the east and maintains an open-water zone to the west. The iceberg-seafloor interaction is currently highly simplified in the model, and the development of a more comprehensive representation in collaboration with our British colleagues (Kostov et al., 2025) now allows us to study icebergs’ grounding time as well as their response to climate change.

J. Petit and P. Mathiot developed the Stand-Alone-Berg (SAB) module to improve the computational performance of NEMO’s Lagrangian iceberg module. Instead of using all NEMO resources, the SAB now runs icebergs offline on a reduced and shallower grid, limited to regions and depths where icebergs are present. It is coupled to NEMO using OASIS3-MCT-v5.0. The NEMO-SAB approach was physically validated and shown to significantly speed-up NEMO when run with iceberg representation (see the NEMO-SAB documentation and tutorial for detail).

The full report can be found in the final scientific report (permission needed).