Human mobility is a fundamental aspect of social, economic, and environmental systems, reflecting how individuals and groups navigate space and time. Mobility patterns reveal critical insights into societal connectivity, resource accessibility, and infrastructure use. Disruptions such as natural disasters, pandemics, and wildfires significantly alter these patterns, exposing vulnerabilities and resilience within communities. Despite the increasing availability of fine-grained mobility data, gaps remain in our understanding of how such disruptions influence mobility across different socio-geographical contexts. Furthermore, the complexity and heterogeneity of movement data pose challenges for integrating and analyzing diverse datasets at varying spatial and temporal resolutions.
This dissertation addresses these challenges by developing advanced computational methods to explore and quantify spatio-temporal mobility patterns in the face of disruptive events. Central to this research is the integration of multimodal movement and contextual data, enabling a nuanced analysis of mobility responses. Using data from the COVID-19 pandemic and wildfire events, this work examines how mobility patterns vary across socio-geographical contexts and time scales, offering insights into heterogeneity in human behavior during crises.
Through a combination of geostatistical analysis, visualization tools, and tensor-based methods, this dissertation makes several contributions. First, it evaluates biases and variability in mobility metrics across multiple data sources, providing a robust framework for assessing their reliability during disruptions. Second, it introduces novel visualization techniques to map and analyze dynamic spatio-temporal mobility clusters. Third, it develops models to quantify the impacts of wildfires on mobility, incorporating contextual factors such as land use and road connectivity. Finally, it leverages tensor decomposition to recover missing data and generate fine-grained temporal patterns, enabling a deeper understanding of movement dynamics.
By integrating these approaches, this dissertation advances the field of movement analytics, bridging the gap between data-driven science, exploratory data analysis and geographic information systems and science. It highlights the importance of data integration, scalability, and contextual analysis in understanding and managing mobility disruptions, offering actionable insights for disaster resilience, urban planning, and policy-making.
