Research Interests
Time Series Analysis, Spatio-temporal Forecasting, GeoAI
Spatial Databases, Big Data Systems
Applied Machine Learning, Physics-Informed Neural Networks (PINNs)
Transportation, Urban Computing, Climate and Environmental Modeling
Research Themes
My research in spatial intelligence is organized around the following five themes:
(1) GeoAI & Spatio-temporal Learning
My work in this area focuses on developing machine learning and deep learning methods to capture, represent, and forecast complex spatio-temporal patterns. Current directions include trajectory representation, prediction, imputation and augmentation, anomaly detection, and routing, as well as leveraging large language models (LLMs), geospatial LLMs (GeoLLMs), and geospatial foundation models (GeoFMs) for large-scale spatial and spatio-temporal reasoning and analytics.
Recent work includes self-supervised contrastive learning for trajectory representation and similarity computation [2026 · Preprint].
(2) Spatial Databases & Big Data Systems
This research theme focuses on developing scalable and intelligent data management techniques for large-scale spatial and spatio-temporal workloads. My current work investigates learning-based approaches for improving spatial database operations, including indexing, query processing and optimization, and workload management, as well as building scalable data infrastructures to support GeoAI and large-scale spatial analytics.
Earlier work surveyed existing spatio-temporal data systems [2021 · PDF], benchmarked big spatial data systems [2018 · PDF], and proposed distributed in-memory approaches for scalable spatio-temporal processing [2019 · PDF] [2018 · PDF].
(3) Mobility Modeling & Intelligent Transportation
This line of research focuses on modeling complex mobility patterns and developing intelligent transportation solutions using spatio-temporal learning, physics-informed modeling, graph-based interaction modeling, and reinforcement learning. My current work emphasizes maritime transportation, addressing trajectory forecasting, interaction-aware modeling, collision-risk assessment, and route optimization, while aiming to extend these approaches to other transportation modes, such as autonomous vehicles, UAVs, and aviation.
Recent work has advanced trajectory forecasting for navigational safety, including physics-informed (PINN) [2026 · Preprint] [2025 · PDF], interaction-aware [2025 · Preprint], and clustering-based [2024 · PDF] [2022 · PDF] approaches. Other studies also explore reinforcement learning-based approaches for maritime route planning and optimization [2026 · Preprint] [2025 · PDF], as well as Markovian modeling of maritime navigation behavior [2025 · PDF].
(4) Urban Intelligence
Urban computing leverages data generated across urban spaces to understand and address urban challenges, such as traffic congestion, air pollution, energy consumption, and urban expansion. My research in this area focuses on combining spatial analytics and artificial intelligence to better understand urban dynamics and support planning and decision-making.
Urban Growth & Planning Analytics. Studying urban growth and sprawl, alignment with land-use and zoning plans, and spatial interactions among urban areas undergoing development.
LLMs for Urban Computing. Evaluating the ability of LLMs to reason over heterogeneous spatial and spatio-temporal urban data and reliably address complex urban computing questions.
(5) Climate & Environmental Modeling
Climate and environmental modeling uses observations, climate projections, and computational methods to understand environmental change and its impacts across space and time. My research focuses on spatio-temporal climate analytics, localized climate information, and AI-driven modeling to support climate adaptation and resilience.
Climate Adaptation & Resilience. Translating large-scale climate projections into localized and accessible information for communities, planners, governments, and other stakeholders.
Climate Impacts & Community Risk. Investigating how climate-related hazards, including sea-level rise and coastal flooding, affect communities and their vulnerability over time.
AI for Climate Modeling & Downscaling. Developing ML/DL approaches for high-resolution climate modeling and downscaling while preserving important spatio-temporal patterns.
Recent work includes a transferable dual-stream learning approach for high-resolution sea surface temperature downscaling that preserves mesoscale spatial structures [2026 · Preprint].