Abstract: Port development ? as key aspect of sustained economic growth ? implies large investments in coastal structures, increasing the risk of structural failures and operational stoppages due to coastal hazards. Then, this paper presents a new approach to statistically characterize and simulate the long-term wave climate to be applied in port risk assessment methodologies in a context of uncertainty. As a starting-point, the weather types framework is used, relating the multivariate and non-stationary climate conditions in an ocean site (the so-called predictand) to continuous-time Markov process of large-scale sea-level pressure and sea-level pressure gradient fields (the so-called predictor). Based on the predictor?predictand relationship, a comprehensive framework composed by three interlinking stochastic models is developed. Firstly, a statistical model describing the predictor is introduced. It is based on identifying representative daily mean synoptic circulation patterns (the so-called Weather Types, WTs) and its probabilistic chronology model at monthly time-scale. Secondly, a multivariate statistical model for the daily predictand is characterized by Gaussian copulas at WT-scale. Thirdly, a multivariate statistical model for downscaling predictand variables to hourly time-scale from daily data is developed. The results highlight that the proposed methodology is a powerful tool for coastal structures applications because disruptive events inducing reliability, functionality and operability failures can be studied simultaneously. The proposed climate-emulator models the long-term wave climate at hourly time-resolution but with a properly and accurate extreme value distribution, taking into account that the local wave climate variables are cross-correlated (4-dimensional compound events), auto-correlated (at daily time-scale) and non-stationary (intra-annual variability). The model is applied at a location in the South West of the Mediterranean basin for emulating wave and storm surge drivers. A cross-validation procedure is applied to verify accuracy of the model, showing a good agreement of the simulated data with the validation samples

Autoría

Descargar referencia