Content area
Abstract
This dissertation confronts the dual crises of escalating multi-hazard risk and the systemic injustice that dictates the distribution of disaster impacts across communities in the United States. Traditional resilience planning, often focused on single hazards and aggregate economic efficiency, is increasingly insufficient and can inadvertently perpetuate the social vulnerabilities it should mitigate. This research argues for and develops a new paradigm of justice-centered resilience planning, powered by advanced optimization frameworks designed to navigate the complex, multi-objective, and multi-stakeholder realities of disaster management.
To address this challenge, this dissertation develops and validates several interconnected, state-of-the-art analytical frameworks. First, a multi-objective optimization model is introduced that operationalizes distributive fairness as a primary objective using the decomposable Theil index. This allows for the explicit, quantitative analysis of trade-offs between minimizing economic losses, population dislocation, repair times, and ensuring the just allocation of resources. Second, a bi-level optimization structure is formulated to capture the hierarchical leader-follower dynamic between public policymakers and private homeowners, enabling the design of realistic, incentive-compatible mitigation policies. Third, a two-stage stochastic programming model formally links pre-disaster mitigation investments (first-stage) with post-disaster evacuation logistics (second-stage) to derive robust strategies that perform well under deep uncertainty about future hazard scenarios. Finally, a hybrid approach integrating machine learning is developed to serve as a computationally efficient surrogate for complex optimization models, bridging the gap between predictive and prescriptive analytics to enable scalable, building-level decision support.
These frameworks are rigorously applied and validated through comprehensive case studies of two high-risk, socio-economically diverse communities: Lumberton, North Carolina (recurrent flooding) and Seaside, Oregon (compound earthquake-tsunami). The results quantitatively demonstrate significant, non-linear trade-offs between achieving mitigation efficiency and ensuring social justice, identifying critical budget thresholds where community-level investments yield the most significant societal benefit. The models successfully identify optimal, adaptive strategies that significantly reduce projected economic and social impacts while maintaining the fair distribution of resources across vulnerable demographic groups.
The primary contribution of this dissertation is a holistic, operational, and data-driven analytical framework that provides decision-makers with the tools to design and implement more effective and just resilience strategies. By explicitly connecting technical efficiency with social responsibility, this work offers a new foundation for forging communities that are not only more robust in the face of natural hazards but also more just and sustainable for all their residents.
Details
