ORCID

https://orcid.org/0000-0001-5939-1569

Year

2024

Season

Summer

Paper Type

Master's Thesis

College

College of Computing, Engineering & Construction

Degree Name

Master of Science in Civil Engineering (MSCE)

Department

Engineering

NACO controlled Corporate Body

University of North Florida. School of Engineering

First Advisor

Dr. Ramin Shabanpour

Second Advisor

Dr. Thobias Sando

Third Advisor

Dr. Ryan Shamet

Department Chair

Alan Harris

College Dean

Dr. William Klostermeyer

Abstract

Florida encounters multiple natural disasters every year, making evacuation a critical mitigation plan when severe impacts are predicted. Evacuation is a complex procedure that involves everything from household decision-making to large-scale traffic network analysis. It also addresses logistical issues, such as transporting impacted people to shelters or medical centers. Access to shelters, gas stations, and lodgings is of utmost importance, as numerous studies and surveys have shown that some people prefer not to evacuate, even if they perceive the disaster risk to be very high.

Based on this, the present study follows two phases to address gaps in evacuation studies. The first path involves a spatial analysis of Florida to assess accessibility to shelters, gas stations, and hotels from community units (block groups). In this phase, the potential clusters of accessibility across different parts of Florida were analyzed using the Global and Local Moran's I methods. To investigate the simultaneous effects of access and hurricane risks, the Bivariate Local Indicator of Spatial Autocorrelation method is used.

In the second phase, public risk perception of wind and flood during hurricanes is explored. People's perception of risk directly influences decisions such as whether to evacuate and how far to evacuate. This phase was based on several parameters, partially provided by survey results conducted by the University of Berkeley in 2017 during Hurricane Irma. These were enriched with spatial variables extracted from the Department of Transportation (DOT) and other departments on the socio-economic, demographic, and health status of respondents' residences. Also, the output from the first phase of this study is utilized to assess whether the access index influences public risk perception. For this analysis, an ordered probit model is used, as the dependent variables were the levels of perceived flood/wind risk, ranging from extremely unlikely to extremely likely.

Results from the first phase revealed significant areas with high access and high hurricane risk (HH), high access and low hurricane risk (HL), low access and high hurricane risk (LH), and low access and low risk (LL). Further study on the socio-economic background of these clusters highlighted significant differences in household car ownership, with people in LH areas likely not owning cars. This social inequity is also reflected in the Gini index, where LH areas have higher values.

The second phase results showed the significant impact of spatial parameters, such as access to lodgings and area resiliency, on how people perceive the risk of hurricane flooding. This phase also highlighted the importance of social media in shaping public risk perception across all types of evacuation orders, including mandatory, voluntary, and shelter-in-place orders.

The results of this study assist policymakers in two ways: first, by identifying areas with higher needs during evacuation to avoid fatalities and other impacts of large-scale disasters; and second, by demonstrating how internal and external factors influence people's risk perception. This information can be crucial for designing more effective evacuation strategies and communication plans.

Available for download on Wednesday, August 05, 2026

Share

COinS