Spatial Modeling of Tornadoes
Steven D. Jones
Advisor: Dr. Susan L. Cutter
Tornadoes are among nature's most powerful and devastating natural phenomena, and affect the United States more than any other country in the world. For nearly a century, annual tabulations of tornadoes and research on them have been conducted, beginning with maps charting the risk of tornadoes to more sophisticated analyses of tornado climatology. This research provides a comprehensive look into tornado occurrence and probabilities in the continental United States for the timer period 1953-2001. Three questions drive this research: 1) What is the potential risk of tornadoes within the contiguous US?; 2) Do areas with high occurrence values (9ncidence and density) have high probability values?; and 3) What modeling methods are most appropriate for the kinds of tornado data now available?
This research used many techniques found in previous assessments, but enhanced them using GIS-based spatial modeling procedures. Three methods (density, occurrence, and probability) were employed using the 49-year dataset. Each of these methods was employed at three different scales (25 by 25 mile grid, 50 by 50 mile grid, and 100 by 100 mile grid) to improve the spatial representation of tornado risks. Probability methods are the most ideal for risk visualization, however, the lack of accurate data (longitude and latitude of touchdown and lift off, path width, path length) limits the effectiveness of this method. Incidence and density models (which require only point locations) are recommended for a quick overview of tornado geography for the past half-century. The different spatial resolutions benefit a wide range of users from planners, to regional climatologists, to the media, and help in visualizing tornado risks.