Prediction and logistics models have yet to be integrated for hurricane preparedness and response. The state-of-the-art generates a point estimate, representing an “average” or “worst-case” scenario for the hurricane’s impact, which becomes input to make preparedness and response decisions in a decoupled fashion. For patient evacuation, this means allocation and routing recommendations for each hospital, ambulance and patient are made independently. We first plan to generate an ensemble of flood scenarios using the state-of-the-art predictive earth, water and atmosphere modeling by taking advantage of high-performance computing and probabilistic nature of hurricane forecasts, including path, wind speed, size and precipitation amounts. We then propose a scenario-based stochastic model that optimizes all evacuation decisions simultaneously, avoiding the drawbacks of sequential, suboptimal decisions. Overall, we seek the best ways to integrate geosciences modeling with hurricane response logistics modeling.
Erhan Kutanoglu (Mechanical Engineering), Zong-Liang Yang (JGS), and John Hasenbein (Mechanical Engineering)