Wind-wave prediction equations for probabilistic offshore hurricane hazard analysis

The evaluation of natural catastrophe risk to structures often includes consideration of uncertainty in predictions of some measure of the intensity of the hazard caused by the catastrophe. For example, in the well-established method of probabilistic seismic hazard analysis, uncertainty in the intensity measure for the ground motion is considered through so-called ground motion prediction equations, which predict ground motion intensity and uncertainty as a function of earthquake characteristics. An analogous method for evaluating hurricane risk to offshore structures, referred to herein as probabilistic offshore hurricane hazard analysis, has not been studied extensively, and analogous equations do not exist to predict offshore hurricane wind and wave intensity and uncertainty as a function of hurricane characteristics. Such equations, termed here as wind and wave prediction equations (WWPEs), are developed in this paper by comparing wind and wave estimates from parametric models with corresponding measurements during historical hurricanes from 22 offshore buoys maintained as part of the National Data Buoy Center and located near the US Atlantic and Gulf of Mexico coasts. The considered buoys include observations from 27 historical hurricanes spanning from 1999 to 2012. The 27 hurricanes are characterized by their eye position, translation speed, central pressure, radius to maximum winds, maximum wind speed, Holland B parameter and direction. Most of these parameters are provided for historical hurricanes by the National Hurricane Center's H*Wind program. The exception is the Holland B parameter, which is calculated using a best-fit procedure based on H*Wind's surface wind reanalyses. The formulation of the WWPEs is based on two parametric models: the Holland model to estimate hurricane winds and Young's model to estimate hurricane-induced waves. Model predictions are made for the 27 considered historical hurricanes, and bias and uncertainty of these predictions are characterized by comparing predictions with measurements from buoys. The significance of including uncertainty in the WWPEs is evaluated by applying the WWPEs to a 100,000-year stochastic catalog of synthetic hurricanes at three locations near the US Atlantic coast. The limitations of this approach and remaining work are also discussed.