Ribes et al. (2025) Towards annual updating of forced warming to date and constrained climate projections

Identification

• Journal: Nature Communications
• Year: 2025
• Date: 2025-10-17
• Authors: Aurélien Ribes, Octave Tessiot, Piers Forster, Nathan P. Gillett, Valérie Masson‐Delmotte, Joeri Rogelj, R. Vautard, Tristram Walsh
• DOI: 10.1038/s41467-025-63026-9

Research Groups

• Météo France, CNRS, Univ. Toulouse, CNRM, Toulouse, France
• Direction de la Climatologie et des Services Climatiques, Météo-France, Toulouse, France
• Priestley Centre for Climate Futures, University of Leeds, Leeds, UK
• CCCma, Environment and Climate Change Canada, Victoria, BC, Canada
• LSCE (UMR 8212 CEA-CNRS-UVSQ), Institut Pierre Simon Laplace, Université Paris Saclay, Gif-sur-Yvette, France
• Centre for Environmental Policy and Grantham Institute – Climate Change and Environment, Imperial College London, London, UK
• Energy, Climate and Environment Program, International Institute for Applied Systems Analysis, Laxenburg, Austria
• Institut Pierre-Simon Laplace, CNRS, Université Paris-Saclay, Sorbonne Université, Paris, France
• Environmental Change Institute, University of Oxford, Oxford, UK

Short Summary

This paper demonstrates that annually updating estimates of current human-induced warming and observationally constrained climate projections significantly improves their accuracy. It shows that a 1-year estimate of forced warming is more accurate and stable than a 10-year average for characterizing the current state of the climate system.

Objective

• To assess the robustness of the estimated forced warming to date as an indicator of the state of the climate system.
• To determine if long-term climate projections should be regularly updated by incorporating the latest available observations.
• To investigate the influence of year-to-year internal variability on the stability and accuracy of these indicators at global and regional scales.

Study Configuration

• Spatial Scale: Global mean surface temperature (GST) and regional (mainland France).
• Temporal Scale: Historical period (1850-present), current year estimates, and future projections up to 2100. The study specifically investigates the impact of annual updates.

Methodology and Data

• Models used:
  • Kriging for Climate Change (KCC) statistical method, a Bayesian estimation approach using a Kalman filter to combine model data and observations.
  • Coupled Model Intercomparison Project Phase 6 (CMIP6) models (45 for global, 27 for regional analyses) for historical and SSP2-4.5 scenario simulations.
  • CMIP5 models (31) for RCP4.5 scenario (for retrospective analysis).
  • CMIP3 models (23) for A1B scenario (for retrospective analysis).
• Data sources:
  • Observed global temperature data: Infilled HadCRUT5 dataset (annual and global means, 1850-2023), including earlier versions (HadCRUT3v, HadCRUT4, HadCRUT4-CW) for retrospective analysis.
  • Observed regional temperature data (France): Monthly temperatures over mainland France since 1899 from Météo-France (average of 30 homogenised stations).
  • Pseudo-observations from climate model simulations for perfect model framework evaluation.

Main Results

• The 1-year estimate of forced warming is significantly more accurate than the 10-year estimate for characterizing current forced warming, exhibiting a bias of -0.02 °C compared to -0.13 °C for the 10-year average, with similar standard deviations (0.044 °C vs 0.045 °C) in a perfect model framework.
• Incorporating observations from every new year continuously refines future climate projections, leading to a modest but consistent reduction in uncertainty ranges (e.g., 1.5% shrinkage of confidence interval for 2100 warming per additional year).
• The estimated forced warming for the current year (1-year estimate) demonstrates sufficient stability to be a robust indicator of the climate system's state, without major spurious year-to-year variability.
• Regional scale analyses yield qualitatively similar results, supporting the use of 1-year estimates for current regional warming, although convergence is slower due to a lower signal-to-noise ratio.
• Revisions in underlying observed datasets and climate model generations can induce non-negligible variations (up to 0.1 °C) in warming estimates, but both 1-year and 10-year estimates are equally sensitive to these input data changes.

Contributions

• Provides robust statistical evidence supporting the annual updating of current forced warming estimates and observationally constrained future climate projections.
• Establishes the superior accuracy and stability of a 1-year estimate of forced warming compared to a 10-year average, offering a more timely and relevant indicator for climate monitoring and policy.
• Validates the Kriging for Climate Change (KCC) method as a reliable tool for providing consistent, seamless estimates of past, present, and future warming, accounting for model uncertainty without requiring frequent climate model reruns.
• Highlights the value of integrating the latest observations for providing up-to-date, policy-relevant information on climate change, aiding authors of future IPCC reports and complementing other climate indicators.

Funding

• Agence Nationale de la Recherche–France 2030, PEPR TRACCS programme (Grant ANR-22-EXTR-0005)
• Météo-France
• CNRS
• World Climate Research Programme (WCRP) through its Working Group on Coupled Modelling (CMIP6 coordination)
• Multiple funding agencies supporting CMIP6 and ESGF

Citation

@article{Ribes2025Towards,
  author = {Ribes, Aurélien and Tessiot, Octave and Forster, Piers and Gillett, Nathan P. and Masson‐Delmotte, Valérie and Rogelj, Joeri and Vautard, R. and Walsh, Tristram},
  title = {Towards annual updating of forced warming to date and constrained climate projections},
  journal = {Nature Communications},
  year = {2025},
  doi = {10.1038/s41467-025-63026-9},
  url = {https://doi.org/10.1038/s41467-025-63026-9}
}