Combination of Internal Variability and Forced Response Reconciles Observed 2023 Warming

The record-breaking global mean surface temperature (GMST) in 2023 came as a surprise to the scientific community, raising the question whether 2023 provides evidence for a recent and abrupt increase in the global warming rate. Here, we quantify the variability and forced contribution to annual GMST by training a statistical learning model on surface temperature anomalies from historical and future climate model simulations. Our method presents a novel approach to separting variability from forced changes in GMST, providing a computationally simple and powerful alternative to existing atmospheric Green’s function approaches. We find that more than half of the 2023 jump in GMST is explained by internal variability, largely owing to anomalously cool conditions in 2022. An unlikely combination of strong but not unprecedented forced and internal contributions occurring simultaneously appears to have led to the extreme jump in 2023 GMST, with North Atlantic warming being a key contributor. When adjusting for variability, we find a steady increase in forced warming rate over the past decades, consistent with previous studies. There is insufficient evidence for an acceleration of forced warming in 2023 and 2024 beyond the expected increase from continued carbon dioxide emissions and decreasing aerosol forcing in the past decades. Our results highlight the role of internal variability for short-term GMST fluctuations and call for an improved understanding of the Atlantic warming observed in 2023.