Modeling multiscalar influences on natural hazards vulnerability: a proof of concept using coastal hazards in Sarasota County, Florida

Abstract

Quantitative vulnerability modeling provides local decision makers and planners with tangible scores that they can utilize to develop and implement effective mitigation strategies. Considering appropriate scale and quantitative methods in vulnerability assessments can help decision makers develop hazard mitigation policies that are effective at the jurisdictional level at which they are implemented. Several types of statistical approaches such as principal component analysis, classical regression and simultaneous autoregressive models have been used to measure vulnerability at various scales (e.g. state or county), but these approaches have limitations for measuring sub-county vulnerability. This paper discusses existing statistical methods utilized in vulnerability assessments and presents a hierarchical generalized linear regression model (Hierarchical GLM) with multiscalar indicators and spatial components. Our model results illustrate that certain indicators are spatially and scalar dependent, providing evidence that examining vulnerability at a sub-county scale while also maintaining a multiscalar perspective offers additional information about social-environmental vulnerability than provided by typical approaches.