What we do
We connect physical hazard simulation, earth observation, human behavior, and decision support to understand disaster impacts and improve how communities prepare and respond.
Research architecture
Our research spans four connected themes: tsunami forecasting, agent-based modeling, Earth observation, and disaster Digital Twins. AI, machine learning, deep learning, reinforcement learning, large language models, scientific computing, and data fusion are used as appropriate to the research question.
Fast physical simulation, inundation forecasting, uncertainty exploration, and damage estimation for tsunami science and emergency decisions.
Representing people, groups, mobility, decisions, and interactions to study evacuation behavior, congestion, guidance, and planning.
Combining satellite observations, physical knowledge, and geospatial data to assess impacts and support disaster response.
Digital Twins connect these streams through shared representations of hazards, infrastructure, people, observations, and decisions.
Explore the Digital Twin framework →Our distinctive contribution
A satellite image can show where damage occurred; a physical model can estimate where water may go; an agent-based model can test how people respond; and a Digital Twin can bring these layers together for planning and action.
How the themes connect
Real-time sensors, Earth observation, and human mobility.
Hazards, impacts and damage, evacuation, and human behavior.
Images, routes, policies, guidance, bottlenecks, and vulnerabilities.
Maps, reports, training, planning support, and real-time decisions.
Research principles
Models are tested, compared, and documented with explicit assumptions and limitations.
Outputs are designed for planning questions: timing, bottlenecks, exposure, and intervention.
Open workflows, case studies, manuals, and software help others adapt the work.