Hurricane Decision-Making: Risk Tolerance and the Safety Margin Forecast

When hurricanes threaten, making decisions that lead to the safety and well-being of thousands of people is a difficult task. These decisions must be timely and proportional in nature. Emergency Managers often use decision trees relying on the National Weather Service (NWS) to provide decision support at critical moments. They need briefings on the latest forecast, but also require information which factors in uncertainty relative to each of the associated hurricane hazards. In order to make effective decisions having an appropriate safety margin, they need an expression of the most likely scenario along with an expression of the reasonable worst case scenario. Absent this, emergency managers apply empirical rules of thumb adding a category of wind or an additional few feet of storm surge to the forecast according to a declared tolerance for risk. In short, enacted decisions which have safety in mind are predicated on the extent to which preparations should be made and the earliest time in which they should be completed. To objectivize and streamline the process of accommodating the risk tolerance of various decision-makers, probability information for both hurricane-related wind and surge have been made available to forecasters. They are provided in cumulative form and incremental form to address both total risk and timing decisions. Using such probability information can be intimidating, even for more experienced forecasters and decision-makers. This paper will explore how hurricane hazard probability information can be used explicitly for supporting sophisticated decision-making and implicitly for the less sophisticated, whether for a community at large or a specific site. It will address quantitative vs. qualitative means for yielding a reasonable worst case scenario according to critical thresholds, and methods for conveying threat trends for identified locations. Forecaster tools and associated logic will also be presented. These tools enable forecasters to dynamically engage exceedance probability data for optimized threat depictions. Finally, challenges related to subsequent public messaging will be outlined with initial considerations from ongoing social science studies.