Accelerating increase in the duration of heatwaves under global warming

Heatwaves are expected to both increase in frequency and duration under global warming. The probability distributions of heatwave durations are shaped by day-to-day correlations in temperature and so cannot be simply inferred from changes in the probabilities of daily temperature extremes. Here we show from statistical analysis of global historical and projected temperature data that changes in long-duration heatwaves increase nonlinearly with temperature. Specifically, from analysis informed by theory for autocorrelated fluctuations applied to European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5) reanalysis and Coupled Model Intercomparison Project Phase 6 (CMIP6) climate model simulations, we find that the nonlinearity results in acceleration of the rate increase with warming; that is, each increment of regional time-averaged warming increases the characteristic duration scale of long heatwaves more than the previous increment. We show that the curve for this acceleration can be approximately collapsed onto a single dependence across regions by normalizing by local temperature variability. Projections of future change can thus be compared to observations of recent change over part of their range, which supports the near-future-projected acceleration. We also find that the longest, most uncommon heatwaves for a given region have the greatest increase in likelihood, yielding a compounding source of nonlinear impacts.