OPTIMIZATION APPROACHES FOR CLIMATE-INDUCED DISPLACEMENT: A MULTI-MODEL FRAMEWORK FOR LONG-TERM PLANNING

Abstract

Climate change is reshaping global patterns of human mobility as environmental degradation,prolonged droughts, sea-level rise, and other slow-onset changes gradually push communities to leave their homes. Unlike sudden disasters, these changes build over time, often leading to severe and sometimes irreversible declines in the long-term habitability of these regions. As climate-related displacement increases in scale and frequency, there is a growing need for long-term relocation planning that can address its complex and evolving impacts. This dissertation addresses the long-term climate-induced relocation challenge through three optimization models, each designed to capture a distinct and critical dimension of planned migration. Together, these models offer a comprehensive framework for designing relocation strategies that are proactive, equitable, and responsive to practical constraints and real-world complexities. The first model introduces a temporal network-based approach to resettlement planning thatconsiders evolving demands, fluctuating capacities, and the importance of long-term integration. By decomposing the relocation timeline into distinct intervals, this model constructs a series of time-layered networks that capture the dynamic evolution of relocation systems. Each temporal layer involves solving a minimum-cost flow problem based on cultural distance, aiming to enhance cultural integration over time. The model provides decision-makers with tools to delay or advance flows based on changing conditions, enabling more adaptive and socially responsive resettlement strategies. The second study proposes a multi-commodity relocation framework that categorizes displacedpopulations by their vulnerability levels. Each vulnerability group is treated as a distinct commodity within an optimization structure that simultaneously considers individuals preferences and capacity constraint. The model supports fairness and adaptability by accounting for changes in host country capacities over time, considering individual relocation preferences in planning decisions, and ensuring continued access for the most vulnerable populations. Finally, the third model presents a digital twin inspired framework for long-term relocation planning under evolving conditions driven by climate stress. It simulates changes in population distribution and destination capacity, updating decisions over time. It enhances adaptability by embedding forecasting in the planning process, enabling more effective responses to future uncertainty. Together, these models contribute to the growing body of climate adaptation research by offeringscalable, optimization-based frameworks that address both logistical and social dimensions of planned relocation. Computational experiments conducted across multiple scenarios demonstrate the models’ capacity to balance competing objectives such as maximizing integration, ensuring equitable treatment of vulnerable populations, and optimizing resource use over multiple time periods. These results underscore the importance of data-driven, long-term strategies for mitigating the human impacts of climate change and suggest new directions for future research in humanitarian logistics and other areas of operations research.