Projections Versus Observations of Extreme Temperatures Over Land During 2006–2023

Abstract

Extreme climate events have received increasing attention due to their growing impact on multiple sectors globally. Future projections of extreme temperatures are critical for devising effective mitigation and adaptation strategies. Given that the reliability of climate models in projecting extreme temperatures remains unclear, we conduct a genuine “forecast” verification by comparing the independent near‐term projections of extreme temperature indices over land by the Coupled Model Intercomparison Project Phase 5 (CMIP5) models with the subsequent observations during the out‐of‐sample period of 2006–2023. Results show that, although biases exist, CMIP5 projections capture the main features of the climatological changes of global extreme temperatures about two decades in advance. Moreover, near‐term projections of extreme temperatures appear to be independent of the four considered emissions scenarios. A notable deficiency is identified in the projection of diurnal temperature range (DTR), for which models systematically underestimate the observed increase.