Dissertation Defense - Shiyang Ruan - Department of Geography and Geoinformation Science

PhD Candidate: Shiyang Ruan

PhD of Science, Earth Systems and Geoinformation Sciences
Department of Geography and Geoinformation Science 

Dissertation Title: From Data to Simulation: Scenario Driven Use of Agent-Based Models to Study Urban Mobility Patterns

Dissertation Chair: Dr. Dieter Pfoser (GMU, Geography and Geoinformation Science) 

Committee Members: 

Dr. Nathan Burtch (GMU, Geography and Geoinformation Science) 
Dr.  Andreas Züfle (Emory, Computer Science)
Dr. Hamdi Kavak (GMU, Department of Computational and Data Sciences) 

Abstract: 

As cities continue to grow in size, complexity, and diversity, ensuring equitable, efficient, and resilient mobility systems has become a critical challenge for urban planners and policymakers. Traditional methods for evaluating transit accessibility and forecasting travel behavior often fall short in addressing the dynamic and spatially heterogeneous nature of urban life. This dissertation advances the field of computational urban science by developing scalable, reproducible, and data-driven frameworks that integrate geospatial analytics, agent-based simulation, and behavioral modeling to understand, assess, and simulate urban mobility at multiple spatial and temporal scales.

This dissertation addresses these challenges through three main contributions that collectively advance computational frameworks for assessing and modeling urban mobility.

First, it presents a comprehensive method for identifying "transit deserts", areas where public transportation supply fails to meet the needs of transit-dependent populations. This is achieved by integrating open-source geospatial data (e.g., GTFS, ACS, LODES, OSM) and constructing spatial and temporal indicators that capture transit demand, infrastructure distribution, service frequency, and accessibility. Using both a quartile-based overlap analysis and Bivariate Local Indicators of Spatial Association (BiLISA), the study maps and statistically validates disparities in the Washington, D.C.–Maryland–Virginia (DMV) metropolitan area.

Second, it develops a digital urban mobility twin using the Multi-Agent Transport Simulation (MATSim) platform to simulate behavioral responses to disruptions, with a specific application to the COVID-19 pandemic. This contribution includes the creation of synthetic populations, multimodal transportation networks, and calibration strategies to reflect pandemic-era constraints, thereby offering insights into system resilience and policy evaluation under stress conditions.

Third, it introduces a novel, needs-driven approach to travel demand synthesis using the Patterns of Life (POL) framework. Unlike data-intensive models, this agent-based simulation generates long-term, high-resolution activity schedules based on behavioral theories (e.g., Maslow’s hierarchy of needs), making it a scalable and transferable method for regions lacking detailed mobility data.

Together, these contributions provide a robust and reproducible toolkit for analyzing transit equity, simulating urban mobility under disruption, and generating realistic activity patterns in diverse data environments. The findings support evidence-based planning and offer practical pathways toward more inclusive, adaptive, and sustainable urban mobility systems.