Decoopman et al. (2026) Climatology and trends of extreme precipitation in France: evaluation of an explicit-convection regional climate model

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-09-16
• Authors: Nicolas Decoopman, Juliette Blanchet, Antoine Blanc, Cécile Caillaud
• DOI: 10.5194/hess-30-5833-2026

Research Groups

• Institute of Geosciences of the Environment (IGE), Univ. Grenoble Alpes, CNRS, INRAE, IRD, Grenoble INP, France
• Service Restauration des Terrains en Montagne (RTM), Office National des Forêts (ONF), France
• Centre National de Recherches Météorologiques (CNRM), Météo-France, CNRS, Univ. Toulouse, France

Short Summary

This study evaluates the convection-permitting regional climate model AROME's ability to reproduce the climatology and trends of extreme daily and hourly precipitation in France (1959–2022). It finds the model valuable for daily extremes but highlights significant limitations in accurately representing the magnitude and spatial heterogeneity of convective hourly extremes and their trends.

Objective

• To evaluate the ability of the convection-permitting regional climate model AROME (2.5 km resolution) to reproduce the climatology and trends of extreme precipitation (specifically 10-year return levels) at daily and hourly scales in metropolitan France, compared to Météo-France station observations.

Study Configuration

• Spatial Scale: Metropolitan France, computational domain ALPX-3 (approximately 1700 km × 2100 km), with 87,536 grid cells over metropolitan France. Model resolution: 2.5 km.
• Temporal Scale:
  • AROME simulation period: 1959–2022 (63 years).
  • Daily station data period: 1959–2022.
  • Hourly station data period: 1990–2022.
  • Trends analyzed over: 1959–2022 (daily) and 1990–2022 (hourly).

Methodology and Data

• Models used:
  • CNRM-AROME (AROME): Convection-Permitting Regional Climate Model (CP-RCM) at 2.5 km resolution.
  • Extreme Value Theory (EVT) using Generalized Extreme Value (GEV) distribution for block maxima.
  • Non-stationary GEV models (M1, M2, M3, M1, M2, M3*) with linear and breakpoint trends in location (μ) and scale (σ) parameters.
• Data sources:
  • Météo-France stations: 1583 daily precipitation stations (1959–2022) and 574 hourly precipitation stations (1990–2022).
  • ERA5 global reanalysis: Used for forcing AROME at boundaries (50 km resolution) and for sea surface temperatures (30 km resolution).
  • ALDERA reanalysis: Provided monthly evolving aerosols.

Main Results

• Climatology: AROME accurately reproduces the large-scale spatial structures of daily precipitation climatology (wet days, annual totals, 10-year return levels) with high spatial correlations (r > 0.94). However, it systematically underestimates hourly 10-year return levels (mean error -6.37 mm, -23.30%), particularly in spring and summer, with lower spatial correlation (r = 0.78).
• Daily Trends (1959–2022): Station data show significant increases in the Rhône Valley (+5% to >+30%) and southern Alps (+20% to +30%). AROME reproduces the broad spatial organization of these daily trends but generally underestimates their amplitudes (negative bias).
• Hourly Trends (1990–2022): Station data reveal highly heterogeneous, noisy, and locally extreme trends (up to ±50%, occasionally >+100%), with strong seasonal dependence (e.g., median +101.86% in June). AROME simulates weak and spatially inconsistent hourly signals, failing to capture the magnitude and fine-scale spatial organization of observed trends. Spatial correlations are very low (e.g., r ≈ 0.12 for the hydrological year, near zero for convective months).
• Temperature Trends: AROME's simulated temperature trends are approximately one-third weaker than observed, contributing to the underestimation of precipitation trend magnitudes.

Contributions

• Provides the first comprehensive evaluation of the CNRM-AROME (2.5 km) model's ability to reproduce both the climatology and trends in extreme precipitation (10-year return levels) across metropolitan France.
• Offers the first nationwide characterization and mapping of hourly 10-year return-level trends in France, revealing structured yet highly localized patterns.
• Utilizes a unique 63-year convection-permitting simulation (1959-2022), which is a significant computational achievement.
• Highlights the value of convection-permitting models for daily extreme precipitation studies while underscoring their current limitations for hourly convective extremes.

Funding

• Agence Nationale de la Recherche (PEPR TRACCS program, grant nos. ANR-22-EXTR-0005 and ANR-22-EXTR-0011)
• EU HORIZON EUROPE Climate, Energy and Mobility (IMPETUS4CHANGE project)

Citation

@article{Decoopman2026Climatology,
  author = {Decoopman, Nicolas and Blanchet, Juliette and Blanc, Antoine and Caillaud, Cécile},
  title = {Climatology and trends of extreme precipitation in France: evaluation of an explicit-convection regional climate model},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-5833-2026},
  url = {https://doi.org/10.5194/hess-30-5833-2026}
}