Climate mitigation benefits emerge within a decade

Discernible differences in global climate responses under varying greenhouse gas emission scenarios are commonly assumed to emerge only after 20 to 30 years.  Here we show that mitigation benefits are detectable within a decade (9±6 years) over the global land area when high-resolution gridded climate data are analysed using a machine learning approach. Specifically, we train an ensemble of gradient-boosted decision tree models on CMIP6 simulations to distinguish between low- and intermediate-emissions scenarios (SSP1-2.6 and SSP2-4.5, respectively) using monthly near-surface air temperature fields. By retaining spatial information, we uncover regional warming signals that remain hidden when relying on global averages and identify the regions in which these signals first emerge using an explainability framework. As a performance baseline, we replace the machine learning approach with a logistic regression model using only global mean surface air temperature, which yields emergence timescales of about 30 years, consistent with previous studies. The spatial pattern of the timing of emergence shows pronounced regional contrasts, with the Tropics standing out as the earliest emerging regions. Even when restricting our analysis to subregions, we find a detectable signal to emerge over the land area of the four highest emitting countries in 13 (±6) years. These results demonstrate that detectable climate benefits of greenhouse gas mitigation appear much earlier than previously recognised and suggest that high emitting countries would also experience near-term benefits from bending the emissions curve. Demonstrating that mitigation produces a discernible climate response within a decade provides a clearer scientific basis for maintaining and accelerating ambitious emissions-reduction efforts.