Frank Tatom (Engineering Analysis Inc.)
Various approaches have been developed for providing a realistic measure of tornado probability for a specified region of interest. These approaches can generally be divided into two categories: 1) event-based and 2) area-based. Event-based approaches, which are in common use, generally deal with the probability of a tornadic event occurring in some specified region during some specified time period. These approaches, however, generally cannot provide a realistic, absolute, estimate of the probability of an object, in the region of interest, being struck by a tornado. Area-based approaches are concerned with the ratio of disturbed land area to total land area in the specified region during a specified time interval. Such approaches can provide a realistic estimate of an object in the region of interest being struck by a tornado. For this reason, area-based approaches are inherently more useful than event-based approaches. At least five area-based models are currently available. The earliest of these was developed by Thom in 1963 while working in the Office of Climatology, U.S. Weather Bureau. Markee, Beckerely, and Sanders developed a similar model for the U.S. Atomic Energy Commission in 1974. Schaefer, Kelly, and Abby derived the Minimum Assumption Tornado-Hazard (MATH) probability model in 1988. Five years later, in 1993, based on the MATH model, Tatom derived the Annual Coverage Fraction (ACF) model, which is currently installed in the Site Assessment of Tornado Threat (SATT) software. Most recently, in 2009, the Federal Emergency Management Agency (FEMA) derived a mean-path-area model, which is very similar to the model by Markee et al. The presentation is designed to provide a comparison of the five area-based models.