Substantial changes in the probability of future annual temperature extremes

Extreme temperature events causing significant environmental and humanitarian impacts are expected to increase in frequency and magnitude due to global warming. The latest generation of climate model projections, Coupled Model Intercomparison Project Phase Six (CMIP6), provides a new and improved database to investigate change in future daily scale extreme temperature events. This study examines the changes in 1, 3, and 5 day averaged annual maximum temperature in four large CMIP6 ensembles. It analyses, using a generalized extreme value (GEV) method, the change in extreme daily mean temperatures at 1.5 and 2°C of global warming, levels highlighted by the 2016 Paris Agreement, and additionally at 3°C. Extremely hot events are characterized using the annual maxima of daily near surface air temperature in the SSP370 scenario. Global changes in the mode of the distributions (location parameter) follow long‐term summer warming and show very similar spatial patterns. Changes in variability (scale parameter) show a clear trend of increases over the tropics and decreases over higher latitudes, while changes to the tails of distributions (shape parameter) show less globally consistent trends but clear signals over the Arctic sea ice, behaviour also seen in variability. Risk ratios (RRs) indicating the change in probability of hot daily extremes that currently have a 10 year return period increase globally with mean temperature change, with greater increases over the tropics. Globally averaged changes in RR over land range from 3.1–3.6 to 7.9–8.3 for 1.5 and 3°C of warming, respectively. For the latter case, this indicates previously rare, once‐in‐a‐decade summer extremes will occur almost annually in the future under high warming. Future extreme heat events are expected to increase significantly with global warming, causing far‐reaching humanitarian, environmental and economic damage. We examine the characteristics of future daily scale annual maxima by applying the generalized extreme value distribution to data sets obtained from CMIP6 daily model output at 1.5, 2, and 3°C of warming. Our study shows significant increases in the probabilities of current events globally, with substantial regions seeing current 1‐in‐10 year events annually at higher levels of warming.