Sensitivity of probabilistic storm surge products to forecast track errors for Hurricane Ike

In order to communicate uncertainties in storm surge forecasts, the National Hurricane Center (NHC) issues storm surge probability (p-surge) products when a hurricane watch or warning is in effect for a portion of the coast of the continental United States. There are two forms of the product, each based on the NHC official track and intensity forecasts but incorporating historical track and intensity errors, as well as variability in the radius of maximum winds. The primary p-surge forecast graphics indicate the cumulative probability that a specified storm surge height—available in one-foot increments between two and twenty-five feet—will be met or exceeded at locations along the coast during the next 72 hours (3 days). The second form, the “exceedance height” graphics, indicates the storm surge height that has a certain probability of being exceeded over the next three days. The exceedance height products are already being used by the local National Weather Service Forecast Offices in the creation of storm surge hazard graphics during a tropical cyclone threat, but it is unknown which exceedance height thresholds are appropriate for given forecast situations and time prior to landfall.

The purpose of this study is to investigate the sensitivity of the second form of the p-surge product—the exceedance height—to errors in the NHC forecast track of Hurricane Ike (2008), using a “perfect” forecast based on the best track and hypothetical track forecasts on either side of the perfect forecast. The exceedance height products are generated for each of the track forecasts and are then compared to the water level observations made along the Texas and Louisiana coasts during Ike. However, storm surge verification has historically been hindered by the relative lack and reliability of water level measurements during and after a storm, so the study also compares the exceedance height products to the best track run of the SLOSH model, which serves as ground truth in the absence of sufficient data coverage. Comparisons of the p-surge output are made for each of the hypothetical storms at various lead times before landfall.