Skip to Content

College of Arts & Sciences
Hazards & Vulnerability Research Institute

Place Vulnerability to Tornadoes in the United States: A Multi-Scale Assessment

Jamie Dustin Mitchem

Advisor: Dr. Susan L. Cutter



Prior research on tornadoes has investigated tornado climatology (risk of tornado occurrences, tornado strength, and impacts), societal vulnerability to this hazards, and the effectiveness of various mitigation techniques.  This study is one of the first to combine all three measures to develop a more comprehensive estimate of the vulnerability of places (counties) to these natural events.  Two research questions are posed.  First, what is the spatial variability in place vulnerability to tornadoes, and what factors (social, physical, mitigation) are responsible for this variation> Second, how does place vulnerability operate at different spatial scales (e.g. county level to a particular community)?

A modified physical vulnerability index was created using tornado segment data from 1950-2002 for all US counties.  Social vulnerability indicators were culled from the 2000 Census and were used to create a composite social vulnerability index that included eight separate sub-indicators.  All county-level variables were standardized using the distance from the median value divided by the interquartile range.  A factor analysis (principal components analysis) was used to create vulnerability indices from the standardized variables.  A county mitigation index was constructed from a number of sources (e.g. building codes, StormReady participation) using similar procedures.  The place vulnerability of counties was computed as the sum of physical vulnerability plus social vulnerability minus mitigation.  Case studies of Marion, IN and Georgetown, SC, which each experienced a significant tornado in 2002, were used to compare the national results at a sub-county level.

The national assessment of place vulnerability to tornadoes showed considerable geographic variation.  As expected, most of “Tornado Alley” was identified as highly vulnerable when risk, social characteristics, and mitigation were considered.  However, portions of Tornado Alley (rural parts of Kansas and Nebraska) had low place vulnerability index values.  Other counties in the Deep South, Midwest, Florida, Colorado, and urban counties were also highly vulnerable.  The place vulnerability index significantly correlated with observed tornado losses, but the strength of the relationship was only moderate (r = .27).  Physical indicators and mitigation have more influence on place vulnerability than social indicators.

The case studies provide details on the nature of the vulnerability at a sub-county level and new d3etaisl about perceptions, preparedness, warning dissemination effectiveness and shelter-seeking behavior in response to a tornado event.  Marion County, IN had a higher place vulnerability score than Georgetown County, SC largely due to more physical risk.  Better mitigation offset the risk, but a larger population size and the expanse of the built up area (and enhanced potential for destructive losses)_ added more social vulnerability to Marion County.  The results of the two public surveys provide two important findings that were consistent: older residents were less likely to seek shelter; and African-Americans were less likely to understand the meaning of a tornado watch.

Synthesizing the three main components of place vulnerability provided a more comprehensive view of the most tornado-prone counties in the US.  This knowledge is important in determining how to best develop methods and programs for loss reduction.  Mitigation efforts can become more cost effective by targeting specific socioeconomic groups in those highly vulnerable areas identified in this research.