Seeber et al. (2026) The observed September 2023 temperature jump was nearly impossible under standard anthropogenic forcing

Identification

• Journal: Communications Earth & Environment
• Year: 2026
• Date: 2026-01-10
• Authors: Svenja Seeber, Dominik L. Schumacher, Sonia I. Seneviratne
• DOI: 10.1038/s43247-026-03178-8

Research Groups

Institute for Atmospheric and Climate Science, ETH Zurich, Switzerland.

Short Summary

This study investigates the unprecedented September 2023 global temperature jump of nearly 0.6 °C, finding it virtually impossible under standard anthropogenic forcing but plausible with a 0.1% probability when accounting for additional external drivers like excess shortwave forcing.

Objective

• To determine if the September 2023 global temperature jump was primarily due to internal climate variability or amplified by external forcings, and to assess its predictability under current and future anthropogenic warming levels.

Study Configuration

• Spatial Scale: Global, regional (tropical/extratropical land/ocean), and grid-cell level.
• Temporal Scale: Monthly (September 2023), annual, and multi-decadal to centennial (1850–2100) analyses.

Methodology and Data

• Models used:
  • Coupled Model Intercomparison Project Phase 6 (CMIP6) ensemble (51 models, 348 realizations).
  • Community Earth System Model version 2 Large Ensemble (CESM2-LE, 100 members).
  • Community Earth System Model (CESM 2.2.2) with observationally constrained surface ocean conditions and nudged atmospheric circulation.
• Data sources:
  • Satellite/Observation: GISTEMPv4 (Global Mean Temperature), HadCRUT5 (Global Mean Temperature).
  • Reanalysis: ERA5 (2-metre temperature, downwelling shortwave radiation, total column water vapour), ERA5-Land (soil moisture in upper 100 cm).
  • Indices: NINO3.4 index (ENSO).
  • Methodology: Non-stationary Generalized Extreme Value (GEV) distribution, Markov Chain Monte Carlo (MCMC) sampler for uncertainty, multiple linear regression for driver quantification.

Main Results

• The September 2023 global mean temperature (GMT) jump of nearly 0.6 °C (0.59 °C GISTEMP, 0.58 °C Copernicus, 0.56 °C HadCRUT) from September 2022 was statistically extreme and virtually impossible under standard anthropogenic forcing, with observation-based probabilities near zero (e.g., GISTEMP: 0 [0–10⁻²], HadCRUT: 10⁻⁴ [0–10⁻³]).
• CMIP6 models also show vanishingly small likelihoods for this jump at the 2023 global warming level (median 10⁻⁶).
• The probability of such a jump increases to 0.1% when accounting for additional external drivers, specifically an estimated excess shortwave (SW) forcing of approximately 0.07 °C.
• The heat was disproportionately concentrated over land, particularly in the extratropics, which accounted for more than half of the observed temperature jump despite covering only one-third of Earth's surface.
• The event resulted from a complex interplay of drivers, including unusually high shortwave forcing, amplified by water vapour feedback, the onset of El Niño, and declining soil moisture.
• An observationally constrained CESM2 simulation closely reproduced the land warming pattern and the magnitude of the temperature jump (0.56 °C vs. 0.58 °C ERA5), suggesting models can reproduce such extremes given realistic boundary conditions, but free-running CESM2-LE simulations did not.
• While the absolute temperature anomaly becomes less unusual under higher warming, the likelihood of such a large year-to-year temperature jump remains extremely low (not exceeding 1% even by the end of the 21st century).
• Projected increases in the frequency and intensity of extreme temperature jumps towards the late 21st century are primarily driven by increased internal variability (93.5%) rather than changes in the warming rate (6.5%).

Contributions

• Provides a comprehensive probabilistic attribution of the September 2023 temperature jump, combining extreme event attribution with a regional and process-based analysis.
• Quantifies the role of internal variability versus external forcings, showing the event was nearly impossible under standard anthropogenic forcing but plausible with additional external drivers (e.g., excess shortwave forcing).
• Identifies the disproportionate contribution of land regions, especially extratropical land, to the temperature jump.
• Highlights the complex interplay of feedbacks and forcings (shortwave forcing, water vapour feedback, ENSO, soil moisture) driving the event.
• Demonstrates that climate models can reproduce the spatial pattern and magnitude of the jump when constrained by observed ocean and atmospheric circulation, but free-running models do not, suggesting a role for external forcing not fully captured by standard CMIP6.
• Projects future probabilities of such jumps, emphasizing the increasing role of internal variability under higher warming levels.

Funding

• European Union’s Horizon 2020 research and innovation programme (XAIDA project, grant agreement 101003469).

Citation

@article{Seeber2026observed,
  author = {Seeber, Svenja and Schumacher, Dominik L. and Gudmundsson, Lukas and Seneviratne, Sonia I.},
  title = {The observed September 2023 temperature jump was nearly impossible under standard anthropogenic forcing},
  journal = {Communications Earth & Environment},
  year = {2026},
  doi = {10.1038/s43247-026-03178-8},
  url = {https://doi.org/10.1038/s43247-026-03178-8}
}