Évin et al. (2026) Uncertainty sources in a large ensemble of hydrological projections: Regional Climate Models and Internal Variability matter

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-02-20
• Authors: Guillaume Évin, Benoit Hingray, Guillaume Thirel, Agnès Ducharne, Laurent Strohmenger, Corre Lola, Flore Tocquer, Jean‐Pierre Vergnes, Jérémie Bonneau, François Colleoni, Lola Corre, Florence Habets, Frédéric Hendrickx, Louis Héraut, Peng Huang, Matthieu Le Lay, Claire Magand, Paola Marson, Céline Monteil, Simon Munier, A. Reverdy, Jean-Michel Soubeyroux, Yoann Robin, Yves Tramblay, Mathieu Vrac, Eric Sauquet
• DOI: 10.5194/hess-30-1023-2026

Research Groups

• Univ. Grenoble Alpes, INRAE, CNRS, IRD, Grenoble INP, IGE, Grenoble, France
• Université Paris-Saclay, INRAE, UR HYCAR, Antony, France
• Univ Toulouse, CNES/IRD/CNRS/INRAE, CESBIO, Toulouse, France
• Sorbonne Université/CNRS/EPHE, METIS-IPSL, Paris, France
• CNRM, Météo-France, CNRS, Université de Toulouse, Toulouse, France
• UMR Espace Dev (Univ. Montpellier, IRD), Montpellier, France
• UR RiverLy, INRAE, Villeurbanne, France
• INSA Lyon, DEEP, UR 7429, Villeurbanne, France
• UMR RECOVER, INRAE, Aix-Marseille University, Le Tholonet, France
• Département Eau Environnement, EDF-DTG, Saint Martin le Vinoux, France
• Geology Laboratory of Ecole Normale Supérieure, Pierre Simon Laplace Research University, CNRS UMR 8538, Paris, France
• Département LNHE, EDF-R&D, 78401 Chatou, France
• OFB, Direction de la recherche et de l’appui scientifique, Nantes, France
• Météo-France, Direction de la Climatologie et des Services Climatiques, Toulouse, France
• LSCE, IPSL, CEA-CNRS-UVSQ, Univ. Paris-Saclay, Gif-sur-Yvette, France
• BRGM – French Geological Survey, Orléans, France

Short Summary

This study quantifies uncertainty sources in a large ensemble of hydrological projections for Metropolitan France using the QUALYPSO method. It finds that low flows are projected to decrease in southern France, with emission scenarios and regional climate models being dominant uncertainty sources, and highlights that internal variability is often as significant as climate change response uncertainty.

Objective

• To characterize projected future changes for different hydrological metrics in France and assess the agreement among modeling chains.
• To quantify the magnitude of internal variability compared to climate change response uncertainty.
• To determine the contributions of scenario and model uncertainties (GCM, RCM, HM) to the climate change response uncertainty.
• To identify the influence of individual models on projections and significant discrepancies between them.
• To analyze how results vary by location and hydroclimatic context across France.

Study Configuration

• Spatial Scale: Metropolitan France, covering 1735 river catchments and 8981 grid points on an 8 km horizontal grid.
• Temporal Scale: Transient projections for the period 1976–2099, with a focus on the end of the century (2071–2099) relative to a reference period (1976–2005).

Methodology and Data

• Models used:
  • Uncertainty Partitioning: QUALYPSO (Quasi-Ergodic Analysis of Climate Projections Using Data Augmentation)
  • Global Climate Models (GCMs): 6 CMIP5 GCMs (e.g., EC-EARTH, IPSL-CM5A-MR, HadGEM2-ES, MPI-ESM-LR, CNRM-CM5, NorESM1-M)
  • Regional Climate Models (RCMs): 9 EUROCORDEX RCMs (e.g., HadREM3-GA7-05, WRF381P, RACMO22E, RCA4, HIRHAM5)
  • Bias Adjustment Model (BAM): ADAMONT
  • Hydrological Models (HMs): 4 HMs (CTRIP, GRSD, ORCHIDEE, SMASH)
  • Reference Evapotranspiration (ET0): Penman-Monteith formula, Hargreaves equation for net radiation.
• Data sources:
  • Climate Projections: Explore2 dataset, a subset of the CMIP5-EUROCORDEX ensemble (36 regional climate projections) for three emission scenarios (RCP2.6, RCP4.5, RCP8.5).
  • Meteorological Reanalysis: SAFRAN for the reference period (1976–2005) and bias adjustment.
  • Hydrological Projections: Explore2 dataset, providing natural river flow simulations for 1735 catchments.

Main Results

• Temperature (RCP8.5, 2071–2099): Median projected increase of 2.5–4.5 °C in winter and 3.5–6 °C in summer. Summer warming is most pronounced in mountainous and southern areas. Inter-model dispersion ranges from approximately 1.5 °C in winter to 4 °C in summer.
• Precipitation (RCP8.5, 2071–2099): Summer precipitation is projected to decrease by 10–40 % (median), with large inter-model dispersion. Winter precipitation mostly increases (median 0 % in South, >30 % in North).
• Reference Evapotranspiration (ET0) (RCP8.5, 2071–2099): Median projected increase of 20–30 %, with large inter-model dispersion (e.g., 5–95% quantile range of 10–40 % in summer).
• Annual Maximum Daily Precipitation (RX1D) (RCP8.5, 2071–2099): Median projected increase of 0–20 % in the South and 20–30 % in northern areas, with considerable inter-model dispersion and disagreement on the sign of change in the South.
• Low Flows (QMNA) (RCP8.5, 2071–2099): Mostly projected to decrease, from 10 % in the North to 45–60 % in the South (median). Projections largely agree on decreases in most southern regions (except the Alps).
• Mean Annual Discharges (QA) (RCP8.5, 2071–2099): Slight increases (<15 %) in the North and decreases (up to 30 %) in the South (median). High inter-model dispersion, with disagreement on the sign of change in many regions.
• High Flows (QJXA) (RCP8.5, 2071–2099): Slight increases (up to 15–30 %) in the North and decreases in the South (median). Large inter-model dispersion and disagreement on the sign of change.
• Internal Variability (IV): Often substantial, frequently of the same order or larger than the Climate Change Response Uncertainty (CCRU) for most indicators, particularly for precipitation and hydrological flows (e.g., >40 % for low and high flows, locally >100 % for low flows).
• Uncertainty Sources (RCP8.5, 2071–2099):
  • Temperature: Scenario uncertainty is the dominant source across all seasons.
  • Precipitation: RCMs and GCMs are major contributors, with scenario uncertainty also significant for summer precipitation.
  • Low Flows: Emission scenarios are dominant in southern France (>40–60 %), RCMs in the East (>20–40 %), and HMs in the North-West and Alps (>20–40 %). GCM uncertainty is lower than RCM uncertainty.
  • Mean Annual Flows: RCMs are dominant in the East (>40–60 %), GCMs are important across most regions, and emission scenarios contribute significantly in the southern half of France (>20–40 %).
  • High Flows: Climate models (GCMs and RCMs) are the predominant contributors, with HM uncertainty more pronounced in snow-dominated regimes.
  • Residual Variability: Moderate to important for precipitation and hydrological indicators (e.g., 20–40 % for mean annual flows, up to >60 % for high flows), suggesting significant interactions between uncertainty sources.

Contributions

• Presents the largest ensemble of hydrological projections derived from regional climate experiments at a country scale (Metropolitan France).
• Utilizes the QUALYPSO method to robustly partition and quantify all uncertainty sources, including internal variability, within an incomplete and unbalanced multi-model, multi-scenario ensemble.
• Provides a comprehensive geographical assessment of projected hydrological changes and their associated uncertainties across diverse hydroclimatic regimes in France.
• Emphasizes the critical and often overlooked role of internal variability, demonstrating its magnitude relative to climate change response uncertainty for hydrological indicators.
• Identifies the specific contributions of individual climate and hydrological models to overall uncertainty, offering insights for future model evaluation and improvement.

Funding

• French Biodiversity Agency (OFB)
• French Ministry of Ecological Transition (MTECT)

Citation

@article{Évin2026Uncertainty,
  author = {Évin, Guillaume and Hingray, Benoit and Thirel, Guillaume and Ducharne, Agnès and Strohmenger, Laurent and Lola, Corre and Tocquer, Flore and Vergnes, Jean‐Pierre and Bonneau, Jérémie and Colleoni, François and Corre, Lola and Habets, Florence and Hendrickx, Frédéric and Héraut, Louis and Huang, Peng and Lay, Matthieu Le and Magand, Claire and Marson, Paola and Monteil, Céline and Munier, Simon and Reverdy, A. and Soubeyroux, Jean-Michel and Robin, Yoann and Tramblay, Yves and Vrac, Mathieu and Sauquet, Eric},
  title = {Uncertainty sources in a large ensemble of hydrological projections: Regional Climate Models and Internal Variability matter},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-1023-2026},
  url = {https://doi.org/10.5194/hess-30-1023-2026}
}