Heat mortality statistics and weather records from the southern United States during the decade of the 1980's are used in an attempt to define an optimum set of temperature thresholds for initiating advisories.
Examination of weather records for the days preceding heat deaths have led to the development of frequency distributions of duration and intensity of heat conditions associated with mortality. These distributions are used to define sets of thresholds for both heat index and minimum ambient temperature. Individual sets are then examined to identify which produce the best level of warning for the least increase in forecaster workload.
This study focuses on the southern United States but a similar approach can be developed for any particular region.