Marson et al. (2026) The Explore2-2022 climate projections dataset for impact studies over France.

Identification

• Journal: Data in Brief
• Year: 2026
• Date: 2026-03-01
• Authors: Paola Marson, Jean-Michel Soubeyroux, Lola Corre, Raphaëlle Samacoïts, Eric Sauquet, Yoann Robin, Mathieu Vrac
• DOI: 10.1016/j.dib.2026.112659

Research Groups

• Direction de la Climatologie et des Services Climatiques, Météo-France, Toulouse, France
• Centre National de Recherche Météorologique, Météo-France, Toulouse, France
• UR RiverLy, INRAE, Villeurbanne, France
• Laboratoire des Sciences du Climat et de l’Environnement, CEA/CNRS/UVSQ, Univ. Paris-Saclay, Institut Pierre Simon Laplace, Gif-sur-Yvette, France

Short Summary

This paper introduces the Explore2-2022 dataset, a new set of bias-corrected regional climate projections for France, sub-sampled from the EURO-CORDEX (EUR11) ensemble and consistent with CMIP6, designed to support impact studies, particularly on water resources, and characterize climate change uncertainties.

Objective

• To produce a new, bias-corrected regional climate projections dataset for mainland France, specifically tailored for climate change impact studies, with a focus on water resources.
• To enable the characterization and partitioning of various sources of uncertainty regarding climate evolution in France, considering different emission scenarios, climate models, and bias correction methods.

Study Configuration

• Spatial Scale: Mainland France, on a Lambert conformal grid (EPSG 27572) with a horizontal resolution of 8 km (143x134 points, 8981 land points).
• Temporal Scale: Daily resolution, covering a historical baseline period (1951-2005) and future transient scenarios (2006-2100) under three Representative Concentration Pathways (RCP 2.6, RCP 4.5, RCP 8.5).

Methodology and Data

• Models used:
  • Global Climate Models (GCMs) and nested Regional Climate Models (RCMs) from the EURO-CORDEX (EUR11) CMIP5 ensemble.
  • Statistical bias correction methods: ADAMONT [6] and CDF-t [7].
• Data sources:
  • Primary data: Climate simulations from the EURO-CORDEX (EUR11) experiment.
  • Reference data for bias correction: SAFRAN reanalysis [8] over France.
  • Consistency constraint: CMIP6 simulations (SSP5-8.5 scenario) for selection criteria.

Main Results

• The Explore2-2022 dataset consists of 53 bias-corrected regional climate simulations, derived from 9 to 17 GCM/RCM pairs (depending on the RCP scenario) and two bias correction methods.
• It provides 10 daily climate variables (e.g., near-surface specific humidity, total precipitation flux, surface downwelling longwave/shortwave radiation, near-surface wind speed, daily maximum/minimum/average near-surface air temperature, snowfall flux, potential evapotranspiration flux).
• The selection process ensures the dataset accurately reproduces the statistical distribution of climate change signals from the full EURO-CORDEX ensemble and maintains consistency with CMIP6 projections, while offering a reduced ensemble size.
• Uncertainty analysis indicates that for temperature projections, the emission scenario is the main contributor, with a non-negligible contribution from climate models (GCMs in coastal areas, RCMs in mountainous areas) in summer. For summer precipitation, RCMs are the main contributor, followed by emission scenarios. The contribution of bias correction methods to uncertainty is significantly smaller.
• Four "narratives" (individual GCM/RCM projections) were selected to represent contrasting future changes in temperature and precipitation under RCP 8.5, aiding in hydrological impact studies.
• Winter precipitation generally shows an increase signal across simulations, while summer precipitation generally shows a decrease signal, with varying robustness and spatial distribution of changes.

Contributions

• Provides a new, updated, and bias-corrected national reference dataset for climate change impact studies over France, particularly for hydrology, succeeding the DRIAS-2020 dataset.
• Offers a thoughtfully sub-sampled EURO-CORDEX (EUR11) ensemble, consistent with CMIP6, which reduces computational burden while preserving the range of projected changes and uncertainties.
• Incorporates two distinct bias-correction methods and continuous time series, enabling robust uncertainty characterization and partitioning across various sources (GCM, RCM, scenario, internal variability, bias correction method).
• Serves as a foundational dataset for constructing the reference warming trajectory for climate change adaptation (TRACC) in France and is actively used by national climate services for adaptation.
• Proposes a "narrative approach" to facilitate user exploration of individual, contrasting climate futures, complementing probabilistic ensemble analysis.
• Adheres to FAIR Data Principles, ensuring transparency, ease of access, and re-usability of the data for researchers and adaptation stakeholders.

Funding

• National Explore2 project
• French Ministry for Ecological Transition
• National Biodiversity Office
• PEPR TRACCS programme
• French Ministry of the Environment

Citation

@article{Marson2026Explore22022,
  author = {Marson, Paola and Soubeyroux, Jean-Michel and Corre, Lola and Samacoïts, Raphaëlle and Sauquet, Eric and Robin, Yoann and Vrac, Mathieu},
  title = {The Explore2-2022 climate projections dataset for impact studies over France.},
  journal = {Data in Brief},
  year = {2026},
  doi = {10.1016/j.dib.2026.112659},
  url = {https://doi.org/10.1016/j.dib.2026.112659}
}