Global mean temperature rapidly warmed during 2023, leading to the warmest year on record at 1.45°C above pre-industrial climate, the causes of which are still under debate. Here we explore the likelihood, mechanisms, and predictability of the rapid warming with the use of a CMIP model suite of large ensemble simulations and a boosted forecast ensemble initialized on Nov 1 2022 with the EC-EARTH fully coupled model run at BSC. We find that CMIP models, on the whole, simulate the observed year-to-year warming between 2022 and 2023 as a 1 in 10000 year event, and 5 out of 8 models never simulate such a large year-to-year increase in global temperature. The forecast ensemblemean predicts about 60% of the observed warming during 2023. The remaining 40% can be accounted for by internally-generated variability by the forecast ensemble, with members that simulate a strong ENSO event during 2023 and positive net shortwave anomalies in nothern latitudes more likely to capture the observed warming. The model also predicts that 2024 will surpass 1.5°C, and forecast members that predict warmer conditions in 2023 also predict a warmer 2024.
Edward Blanchard‐Wrigglesworth (Thu,) studied this question.