As a Jockey Club STEM postdoctoral fellow at the City University of Hong Kong, I am developing an AI-engine framework for terminal airspace optimization that addresses critical challenges in aviation safety, efficiency, and environmental sustainability. My research focuses on accurately generating aircraft trajectories in high-density terminal airspaces where traditional simulation approaches are labor-intensive and existing data-driven methods are limited to areas with historical data. Using both functional data analysis and graph-based spatiotemporal modeling, I extract transferable knowledge from historical trajectory patterns to generate trajectories for new route designs through transfer learning neural networks.

My work embodies the Forum's mission by:
1) Integrating multiple disciplines (data science, aerospace engineering, environmental studies).
2) Translating theoretical models into practical applications with real-world impact.
3) Developing transferable knowledge frameworks that facilitate exchange across scientific communities.

Participating in the Forum would allow me to share methodological insights on modeling complex system uncertainties, learn from Shaw Laureates in related disciplines, establish cross-disciplinary collaborations, and demonstrate to younger scientists how data science can address significant transportation challenges. I am enthusiastic about contributing to scientific advancement in Hong Kong and fostering international collaboration through this distinguished platform.

My research focuses on developing an AI-engine digital framework for terminal airspace optimization and redesign. Terminal airspace surrounding airports features high flight density and complex structure, with aircraft trajectories often deviating from standard routes due to various operational factors. Accurately predicting these trajectory uncertainties is critical for evaluating route redesigns, yet traditional simulation approaches are labor-intensive while existing data-driven methods are limited to airspaces with historical data.

To address this gap, I investigate two approaches for transferable knowledge discovery: functional data analysis and graph-based spatiotemporal modeling. By extracting patterns from historical trajectory behavior, I develop transfer-learning neural networks that can generate trajectories for new route designs without requiring extensive historical data for those specific routes. My integrated evaluation platform assesses multiple aspects of proposed designs including efficiency, safety, feasibility, environmental impact, and noise pollution.

This research will have significant implications for air traffic management, particularly in high-density areas like Hong Kong and the Greater Bay Area. By enabling data-driven evaluation of proposed airspace modifications before implementation, my work contributes to safer, more efficient, and environmentally sustainable aviation operations while developing novel approaches to spatiotemporal modeling with applications beyond the aviation domain.