Extended Data Fig. 2: The return period of record-shattering jumps in sea surface temperature depends sensitively on the warming trend, and the autocorrelation and variance of annual sea surface temperatures.

The return period of record-shattering sea surface temperature (SST) events, defined as events when the annually averaged SST (April to March) between 60°S and 60°N is at least 0.25 °C (record-breaking SST observed in 2023/2024) larger than the previous record SST, is plotted in dependence of the warming trend over a given period and the temperature variability for five different autocorrelations: a) 0.2, b) 0.3, c) 0.4, d) 0.5, and e) 0.6. The return period is calculated based on timeseries of 100 million years with an AR(1) model and prescribed trends, variability and autocorrelation (see methods). The dots indicate each simulation, and their location indicates the trends, variability and autocorrelation derived from these simulations. The larger dots show the simulations that included a record-shattering jump in SSTs. Accordingly, the crosses indicate the trend from NOAA OISST V2.1¹ and variability and autocorrelation from ERA5³⁰ (cross in b)), HadISST³¹ (cross in c)), and ERSST³² (cross in d)).