Low-dimensional offshore wave input for extreme event quantification

In offshore engineering design, nonlinear wave models are often used to propagate stochastic waves from an input boundary the location of structure. Each realization is typically characterized by a high-dimensional time-series, and reliable determination extreme events associated with substantial computational effort. As sea depth decreases, become more difficult evaluate. We here construct low-dimensional characterization candidate time series circumvent search for in probability space. represented unique set parameters which standard surrogate approximations, such as Gaussian processes, can estimate short-term exceedance efficiently accurately. demonstrate advantages new approach simple shallow-water model based on Korteweg–de Vries equation we provide accurate reference solution Monte Carlo method. furthermore apply method fully propagation over sloping seabed. The results that process learn accurately tail heavy-tailed distribution maximum crest elevation only $$1.7\%$$ required evaluations.