Abstract
Summertime temperature extremes exert profound impacts on human health, ecosystems, and infrastructure, yet their characteristics and drivers remain poorly quantified in New Zealand, in part due to the country's complex topography and multiple climate regimes. Using high‐quality temperature observations from 30 weather stations spanning 1980–2022, this study examines the capability of a commonly used global reanalysis product, ERA5, to represent observed characteristics and changes in the frequency of extreme heat over the country during the austral summertime (December–February). Results show ERA5 consistently exhibits a cold bias that is systematically larger (smaller) on the hottest (coolest) days of the summer. As a consequence, ERA5 also consistently underestimates both daily and interannual temperature variability across all stations. Station observations reveal the number of late‐summer (February–March) hot days has increased by up to a factor of three since the 1980s, particularly across northern and eastern New Zealand, despite no consistent long‐term decline in summer rainfall. When examining pseudo‐global‐warming scenarios, increases in the frequency of extreme hot days emerge too rapidly under ERA5, highlighting the importance of correctly representing day‐to‐day variability when examining projected heat extremes in topographically complex regions like New Zealand.