Joint distribution model for prediction of hurricane wind speed and size

This paper suggests a methodology for characterizing the joint distribution of hurricane intensity (maximum wind speed) and size (radius of maximum winds). Such a model represents an extension of traditional wind hazard models by including joint information on the critical spatial dimension. Typically, the hurricane hazard is described in terms of maximum wind speed Vmax (at the eye-wall), since damage descriptors associated with intensity scales (e.g., the Saffir–Simpson Hurricane Scale) and collateral hazards (e.g., hurricane surge) are related most often to maximum wind speed. However, recent studies have shed light on the importance of storm size (i.e., radius of maximum wind, Rmax) in describing the hurricane wind field and thus the spatial extent of potential damage. The large losses from several recent hurricanes underscore the need for better understanding the impact of storm size on damage. To that end, we seek to develop event parameter combinations (e.g., Vmax and Rmax) that define “characteristic” risk-consistent hurricanes in one particular geographic region. A simulation framework is developed to generate 10,000 years of simulated hurricane events and a synthetic hurricane wind speed database for the state of Texas, using state-of-the-art hurricane modeling techniques and information extracted from historical hurricane data. The resulting 10,000 years database, which includes information developed for every zip-code in Texas, includes time of hurricane passage, maximum gradient wind speed and surface wind speed. Using this simulation framework, selected parameters (i.e., intensity and size parameters) are recorded for each hurricane at the time of landfall along the Texas coast. Using a hurricane decay model specifically calibrated for this location, parameters Vmax and Rmax at inland locations also are recorded. The critical values of Vmax and Rmax are then selected to jointly describe the intensity and spatial extent of hurricanes and the joint histogram is developed. Finally, these variables are statistically characterized and a suite of the characteristic Vmax and Rmax combinations corresponding to certain hazard levels are identified. The proposed methodology can be used to develop characteristic hurricane hazard definitions (and event parameter combinations corresponding to specific hazard levels) for use in performance-based engineering applications.

► We suggest a methodology for characterizing the joint distribution of hurricane intensity and size.
► We develop parameter combinations that define risk-consistent hurricanes in one region.
► We develop a simulation framework to construct a synthetic hurricane wind speed database.
► We propose hurricane hazard definitions for use in performance-based engineering applications.