|
Abstract:
|
Extreme wave events are of interest in marine sciences for their relationship to meteorological and sea state predictors as well as the potential impact on the human environment (i.e. coastal erosion, flooding and structural damage). Recent advances in statistical computing and duration modeling offer an opportunity to model wave height in new ways. With regard to the height of waves, one may specify a binary time series of extreme and non-extreme waves of a particular threshold and define a dynamic linear model. Integrated Nested Laplace Approximation (INLA) methods may be used for Bayesian computing offering speed increases over Markov Chain Monte Carlo (MCMC). With regard to the duration between extreme wave events, one may adapt and generalize the family of auto-regressive conditional duration (ACD) models introduced by Engle and Russell. These techniques are applied to public data from the National Oceanic and Atmospheric Administration's National Data Buoy Center.
|
