Abstract Stabilizing global warming under ongoing anthropogenic warming will likely require atmospheric carbon dioxide removal. However, the extent to which carbon dioxide removal can reverse past warming remains uncertain due to divergent responses across a limited set of models. Here, we develop a data-driven emulator based on an energy balance model with three layers that captures the nonlinear temperature response to forcing. Calibrated to state-of-the-art instrumental temperature datasets, our observationally constrained emulator projects that, at the same carbon dioxide concentration, global warming during the carbon dioxide removal phase will be on average 0.7 °C (95% confidence interval: 0.5, 1.0) warmer than during the carbon dioxide increasing phase, narrowing the 0.2–1.9 °C inter-model range by more than 70%. Applied to mitigation scenarios, global warming is projected to temporarily overshoot the 1.5 °C target, even under stringent pathways reaching net-zero carbon dioxide emissions.
Song et al. (Thu,) studied this question.