TexTalks: Disaster Decision Making: Integrated Prediction and Optimization
Title: Disaster Decision Making: Integrated Prediction and Optimization for Resilience Planning
How come we still have devastating weather-related tragedies despite increasingly powerful weather prediction tools? Could it be as simple as “connecting the dots?”
The 10-day power outage following Hurricane Ida in New Orleans resulted in heat-related deaths. Uri, the February ’21 winter storm, caused a power outage that resulted in deaths in Texas due to lack of heating in freezing temperatures.
It is disappointing that we have high-resolution weather prediction models, yet we still struggle to prevent these disasters. One potential culprit is the disconnect between prediction models and disaster decision making. Our research bridges this gap by linking them. When integrated and optimized, our models suggest resilience decisions based on a full grasp of the uncertainty quantified using the scenarios created with the most up-to-date predictions at hand.
Our approach captures the infrastructure network interactions and correlations of potential impacts of the disaster, and explicitly models the dependency between three stages of resilience decision making (mitigation, preparedness and recovery), resulting in well-informed decision all around. We present results from multiple projects using two infrastructures, the healthcare network and the power grid, impacted by two types of events, hurricanes and winter storms (using Harvey and Uri as examples), all tested on realistic Texas-based datasets.