A year above 1.5 °C signals that Earth is most probably within the 20-year period that will reach the Paris Agreement limit

Global mean surface air temperature (GMST) reached 1.43 °C above pre-industrial in 2023, partly influenced by natural variability (including El Niño). In 2024, multiple international datasets indicate Earth’s surface air temperature averaged around 1.55 °C above pre-industrial. However, the Paris Agreement temperature goals are assessed over multi-decadal periods and refer to human-induced warming. The IPCC AR6 uses 20-year periods to estimate human-induced warming and to assess climate risks at specific warming levels (notably 1.5 °C). The central question addressed is whether a single calendar year exceeding 1.5 °C serves as an early warning that the world has entered the first 20-year period whose average warming reaches 1.5 °C, which is critical for anticipating associated climate risks and planning adaptation and mitigation.

The study builds on IPCC assessments (AR6 and the 2023 Synthesis Report) defining warming levels over 20-year periods and differentiating human-induced warming from short-term variability. Prior work has discussed approaches to track progress against temperature goals, uncertainties in reaching thresholds, internal variability influences, and carbon budget uncertainties. The interpretation of temperature goals in geoscience and policy contexts has been clarified by UNFCCC decisions and literature (e.g., Rogelj et al. 2017; UNFCCC Decision 21/CP.27). Recent analyses document 2023–2024 warming anomalies, potential contributions from El Niño, reduced aerosols (e.g., shipping), and volcanic effects. Model selection constraints are informed by Tokarska et al. (2020), ensuring consistency with observed warming trends. The literature emphasizes that threshold crossing can only be confidently established in hindsight due to observational and methodological uncertainties.

Data: Observational GMST anomalies relative to 1850–1900 were taken from Forster et al. (Earth Syst. Sci. Data), including annual series for HadCRUT, NOAA, Berkeley Earth, and Kadow, plus a consolidated 4-set mean. CMIP6 model simulations were used by concatenating historical (1850–2014) with SSP scenarios (SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5). To estimate annual global warming relative to 1850–1900 for models, the observed warming from 1850–1900 to 1995–2014 (−0.86 °C from the consolidated mean) was added to modelled global mean surface air temperature (tas) changes relative to 1995–2014, following IPCC procedures to minimize model bias.

Model ensemble and selection: One ensemble member per model was used (preferably r1i1p1f1 through 2099, otherwise first available with full data). Models were constrained to those with warming trends during 1981–2014 compatible with observations (0.0108–0.0263 °C yr⁻¹; ±2σ range incorporating internal variability, blending, and structural uncertainties per Tokarska et al. 2020). Sensitivity analyses show conclusions are robust without this selection.

Definitions of timing and lag: For each time series and warming level, the entry time of the first 20-year period (t_20yr) was defined as the first year of the earliest 20-year window whose mean warming meets or exceeds the level. The timing of the first single year (t_1yr) was the first annual anomaly meeting or exceeding the level. The lag Δt = t_1yr − t_20yr was computed; Δt ≥ 0 indicates the first single warm year occurred within the relevant 20-year period. If a warming level was never reached, times were set to infinity and such cases handled consistently (e.g., Inf − Inf excluded).

Probability estimation: For climate model ensembles under each SSP, the probability that the first single year at 1.5 °C falls within the 1.5 °C 20-year period was estimated by counting simulations with Δt ≥ 0.

Idealized experiment: To assess dependence on trend and variability, 20,000 synthetic time series (1850–2200) were generated using Gaussian noise with fixed standard deviation and time-varying means representing specified trends. Observed warming from 1850–1900 to 1995–2014 (−0.86 °C) was added to simulated changes relative to 1995–2014. Across a grid of realistic trends and standard deviations (derived from observations and CMIP6), the probability that the first single year at or above a warming level falls within the corresponding 20-year period was computed. Trends were measured over the 20-year period reaching a given warming level; variability was the standard deviation of detrended data.

Trend-reduction analysis: Assuming the first 1.5 °C year occurs under current observed trend and variability, the post-first-year trend was varied to quantify how reducing warming rates affects the probability that this year lies within the 1.5 °C 20-year period. Uncertainty ranges were computed using combinations of the observed trend best estimate and likely (5–95%) range for 2014–2023 (0.026 (0.02, 0.04) °C yr⁻¹) and observed 20-year standard deviation range across datasets (0.084 (0.081, 0.087) °C). Scripts were implemented in R and Bash, with preprocessing via CDO; data and code are openly available.

• Observations show that for warming thresholds from 0.6 to 1.0 °C, the first single year exceeding each threshold consistently occurred within the first 20-year period whose average reached the same threshold. Climate models reproduce this pattern.
• For the 1.5 °C warming level, CMIP6 simulations indicate the first single year at or above 1.5 °C is likely to virtually certain to fall within the first 20-year period reaching 1.5 °C, depending on scenario. The fraction of simulations with Δt ≥ 0: SSP5-8.5 ≈ 96%, SSP3-7.0 ≈ 100%, SSP2-4.5 ≈ 100%, SSP1-2.6 ≈ 73%, SSP1-1.9 ≈ 75%. Under SSP2-4.5 (closest to current policy trends), all selected models indicate Δt ≥ 0.
• Averaged observational datasets indicate 2024 reached ~1.55 °C above pre-industrial, marking the first calendar year above 1.5 °C.
• Strong recent warming rates (best estimate ~0.026 °C yr⁻¹, likely 0.02–0.04 °C yr⁻¹) make it virtually certain that the first 1.5 °C year lies within the 20-year 1.5 °C period. To reduce this probability to ~50%, the warming trend would need to drop to ~0.005 °C yr⁻¹ after the first 1.5 °C year, roughly 20% of the recent decadal rate, requiring stringent mitigation.
• The probability increases with stronger trends and lower interannual variability. Idealized experiments align with model-based probabilities; if model variability is overestimated, actual probabilities may be even higher.
• The model ensemble of opportunity does not represent scenario outcome likelihoods as assessed by the IPCC; probabilities should be interpreted with caution.

The analysis addresses whether a single calendar year above 1.5 °C heralds entry into the first 20-year period averaging 1.5 °C, which is the relevant benchmark for Paris Agreement goals and risk assessments. Both observations and CMIP6 models indicate that, under current warming rates and modest variability, the first 1.5 °C year almost always occurs within the corresponding 20-year period. This provides an early warning that the documented risks associated with 1.5 °C warming begin to emerge within this period, informing adaptation planning and risk management. The findings are robust across observational datasets and supported by an idealized probabilistic framework linking trend and variability to the timing relationship. Caveats include uncertainties in datasets and the representation of recent short-lived forcers and natural variability in SSP forcing trajectories, but these likely render the estimates conservative for scenarios closest to current emissions. The result underscores the urgency of near-term mitigation to slow warming rates, reducing the likelihood and severity of crossing and remaining above the 1.5 °C level, and maintaining pathways to limit peak warming well below 2 °C.

Unless ambitious emissions cuts are implemented rapidly, the first calendar year above 1.5 °C virtually certainly occurs within the first 20-year period averaging 1.5 °C. The strong recent warming trend and relatively low interannual variability explain this high probability. The 2024 year above 1.5 °C most probably signals that Earth has entered the 20-year 1.5 °C period, implying that climate impacts associated with a 1.5 °C world may begin to emerge. This is a call to action: stringent near-term mitigation can markedly lower warming rates, reduce the risk of soon exceeding the 1.5 °C level after the first exceedance, and help keep peak warming well below 2 °C. Future work should refine trend and variability estimates, better incorporate recent short-lived forcer changes into scenario forcing, and improve attribution of single-year anomalies to internal variability versus external forcing.

• The SSP scenario forcings begin in 2015 and may miss recent changes in short-lived climate forcers (e.g., COVID-related emissions, reduced sulfur in shipping fuels), cloud/aerosol effects, and volcanic eruptions, potentially affecting single-year anomalies and near-term trends.
• The ensemble of models used is an ensemble of opportunity and does not represent the full probability distribution of scenario outcomes as in IPCC assessments; scenario-specific likelihoods should be interpreted cautiously.
• Observational uncertainties and dataset differences persist; threshold determinations can only be confirmed confidently in hindsight.
• Climate model interannual variability may be overestimated; if so, model-based probabilities are conservative.
• Idealized experiments assume Gaussian variability and simplified trend structures; real-world processes may deviate.
• Internal variability (e.g., ENSO) can temporarily elevate annual temperatures; attribution between internal variability and forcing for specific years has uncertainty.