Assessment of the Australian Bureau of Meteorology wet bulb globe temperature model using weather station data

Exertional heat illnesses affect thousands of athletes each year and are a leading cause of death in sports. The wet bulb globe temperature (WBGT) is widely used as a heat stress metric in athletics for adjusting activities. The WBGT can be measured on-site with portable sensors, but instrument cost may provide a barrier for usage. Modeling WBGT from weather station data, then, presents an affordable option. Our study compares two WBGT models of varying levels of sophistication: the Australian Bureau of Meteorology (ABM) model which uses only temperature and humidity as inputs and a physically based model by Liljegren that incorporates temperature, humidity, wind speed, and solar radiation in determining WBGT outputs. The setting for the study is 19 University of Georgia Weather Network stations selected from across the state of Georgia, USA, over a 6-year period (2008-2014) during late summer and early fall months. Results show that the ABM model's performance relative to the Liljegren model varies based on time of day and weather conditions. WBGTs from the ABM model are most similar to those from the Liljegren model during midday when the assumption of moderately high sun most frequently occurs. We observed increasingly large positive biases with the ABM model both earlier and later in the day during periods with lower solar radiation. Even during midday, large (≥ 3 °C) underestimates may occur during low wind conditions and overestimates during periods with high cloud cover. Such differences can lead to inaccurate activity modification and pose dangers for athletes either by underestimating heat-related hazards or by imposing an opportunity cost if practice activities are limited by overestimating the heat hazard.