Roads or radar: Investing in infrastructure or improved forecasting in the face of tropical cyclone risk

Given limited funding, should we invest in infrastructure to speed evacuation in an emergency, or in forecasting technology to better predict the timing and intensity of the event? For example, should we build additional evacuation routes along the Gulf Coast of the United States to speed hurricane evacuations or should we improve our ability to forecast the path and intensity of an approaching storm?

In this research, we use dynamic programming to model the evacuation and information-gathering decision of an official responsible for public safety. We assume that at each stage, an evacuation can be ordered that will take several stages to complete, or a decision can be made to wait and gather additional information regarding the approaching storm. Early evacuation mitigates loss, but may ultimately not have been necessary. On the other hand, waiting too long to evacuate could produce tragic consequences.

By integrating the emergency management and forecasting systems into a single dynamic model we are able to determine the interplay and benefit of investments in these systems, thereby allowing the selection of an optimal investment mix.