ABSTRACT
Storm surge is the cause of a significant amount of hurricane damage and all locations along the US East and Gulf coasts are vulnerable to storm surge. Therefore, the prediction of storm surge is an important part of risk analysis for hurricanes. Moreover, although it is still debated, several studies foresee the possibility of climate change increasing the intensity of hurricanes and consequently aggravate the possible damage, including those due to the effects of storm surge. However, the models used for storm surge either are computationally demanding, or have low accuracy and predict storm surge only in locations where the data used to develop the model are available. Consequently, they are unsuited for probabilistic studies that require a large number of simulations, and for timely predictions in the case of approaching hurricanes. A solution is the use of metamodels. This chapter presents a probabilistic model for storm surge built using the combination of a logistic regression and a physics-based random field. Differently from existing metamodels, the presented model can be calibrated not only using data from high-fidelity simulations but also historical records. Moreover, the probabilistic model can predict storm surge in locations different from those where data are available.
