Skip to Content

College of Arts & Sciences
Hazards & Vulnerability Research Institute

Comparing the Vulnerability of the Built Environment Among US Cities

Kevin Allen Borden

Advisor:  Dr. Susan L. Cutter



    This thesis examines the relative vulnerability of the built environment among 132 US cities.  The research questions ask: 1) What factors of the built environment help to define its vulnerability to environmental hazards; 2) Is there regional variability with respect to the vulnerability of the built environment among US cities; and 3) Can a measure of spatial centrality based on the population density within an urban area correlate with a city's level of vulnerability?

    A comparative index of vulnerability was computed with a principal components analysis (PCA) for the built environment using data from the 132 cities in the study area.  Analysis of the results allowed for the identification of the underlying dimensions that contribute to different levels of vulnerability in the built environment.  The built environment vulnerability index (BEVI) produced 11 components explaining 73.6% of the total variance among the input variables.  Some of the components extracted include industry and public utilities, emergency services, landmarks, and transportation hazards.  Mapping the BEVI revealed clusters of high vulnerability in the Northeast, Southeast, and among smaller, non-primary cities.  Lower vulnerability was evident in the Southwest region of the US.

    Spatial autocorrelation, specifically a Moran's I coefficient, was used to define the level of spatial centrality for each city.  The researcher hypothesized that cities with lower levels of centralization would tend to have higher levels of vulnerability.  Correlation tests showed that this was not the case.  Although no statistical relationship was found between this measure of urban form and the vulnerability of the built environment, there was a slight tendency for centralized urban areas to have higher levels of vulnerability.  This was shown in a very weak Spearman's Rho correlation coefficient (r = .240, p = .006), and visually when vulnerability and centrality were mapped.