This study investigates record-shattering temperature extremes using a very large hindcast ensemble, motivated by the record-breaking heat in the Pacific Northwest in June 2021. The hottest days in the ensemble show similar large-scale and synoptic patterns to the observed event. However, these extreme days are acutely sensitive to chance weather patterns, making them rare and requiring extensive samples (tens of thousands of years) to capture their frequency accurately. The persistence of such records is due to the low probability of repeating the optimal weather alignment, even with warming trends. Climate models may not capture such extremes unless their sample sizes are sufficiently large, highlighting the importance of considering sampling effects when assessing models and changes in weather-sensitive extremes.

James S Risbey, Damien B Irving, Dougal T Squire, Richard J Matear, Didier P Monselesan, Michael J Pook, Nandini Ramesh, Doug Richardson, Carly R Tozer