UCLA

Abstract Recent studies have revealed strong trends in humid heat, including the nearing of human physiological limits in some regions. Understanding of past extremes and their meaningfulness for contextualizing future possibilities, especially in the near‐term, is limited by the absence of a global analysis focused on the most extreme humid‐heat‐anomaly events. Here we identify record‐setting humid‐heat days for 216 global regions and assess the likelihood of these records being broken even under present‐day climate forcing. We use several reanalyses as a historical catalogue, and large climate‐model ensembles to represent other statistically plausible events. Unlike the spatial pattern of large temperature anomalies, we find that humid‐heat anomalies are most intense, and most seasonally and interannually concentrated, in the deep tropics and arid subtropics. Many top events have attracted little if any prior attention. The eastern United States is especially susceptible to record‐breaking humid heat due to modest current records (>1% inferred annual exceedance probability) contrasting with numerous simulated large‐anomaly days. Australia and eastern China are also prone to locally exceptional episodes, with >40% of ensemble members simulating events exceeding the ERA5‐based distribution maximum. Model biases for key characteristics, together with the observed record‐setting day affecting its estimated return period by >2.5x in half of regions, underline several valuable aspects of a joint observation/model perspective on humid heat. This approach aids in evaluating the plausibility of as‐yet‐unseen extremes; identifying regions of concern that might otherwise be overlooked and underprepared; and gauging regionally specific correlations between event magnitudes and societal impacts. Plain Language Summary Surprisingly little is known about what the most extreme humid heat ever observed looks like on a regional basis. Understanding these upper limits of past experience would aid in ascertaining how as‐yet‐unseen extremes might affect human and natural systems. Using climate models and observations, we identify areas with the most intense humid heat versus ones with more modest records, and connect those patterns to the El Niño‐Southern Oscillation and other drivers. Our two‐pronged statistical approach highlights the eastern US and much of Australia as particularly susceptible to having their current records broken—in other words, they have a high statistical likelihood of event magnitudes beyond those experienced in recent decades. We also evaluate climate models against reanalyses and identify systematic model biases in humid‐heat magnitude, duration, and propensity for temporal clustering. Our results draw attention to, and aid in quantifying, impactful record‐breaking humid heat, especially in the tropics and subtropics where adaptation efforts should especially be focused. Key Points The largest and most‐clustered humid‐heat anomalies occur in low latitudes, and have not been well captured by in situ networks Model ensembles can identify regions prone to large anomalies, while reanalyses can identify those with modest current record maxima The single record‐setting day often affects its estimated return period by a factor of 3 or more, so good data on extremes is crucial