The impact of projected changes in climate variability on heat stress events in the CESM2 LENS

Humans and other organisms will continue to experience increasing levels of heat stress as the global temperature rises. The warming signal in annual mean temperatures at global and regional scales is robust within the simulated worlds of many climate models, and within the observational record. Changes in variability are far more subtle, requiring higher temporal resolution data and a significant number of initial-condition ensemble members, from individual or multiple climate models, for identification and characterization. As such, the impact of changes in climate variability on heat stress events is not well understood. Here, we characterize heat stress impacts in the 100-member CESM2 Large Ensemble simulation (SSP3-7.0) via 3-hourly Wet Bulb Globe Temperature (WBGT), a widely used standard for determining heat exposure levels by professional, governmental, and athletic organizations. We find that discrete regions including the Persian Gulf, India, and Pakistan are projected to experience the largest number of extremely severe (WBGT>35°C) heat stress events, The regions most impacted by extremely severe heat stress are not the ones that experience the largest warming, indicating that heat waves defined by relative warming may not be suitable for some types of impact assessments. Additionally, these regions are notably absent from observational data used to link temperatures to health impacts. Neglecting WBGT variability changes, and only considering mean-state changes, results in heat stress estimates up to 25% higher or lower than those based on the culmination of mean and variance changes. The impacts of variability changes are regionally heterogeneous. For example, variability changes tend to mitigate the most severe heat stress impacts in Pakistan but increase severe heat stress over most of India. The results imply limits for the accuracy of heat stress projections that do not account for changes in variability.