Metodologías para el análisis de riesgo de inundación en zonas costeras

Abstract

This thesis has been focused on on the development of a novel, robust and statistically rigorous coastal flooding risk assessment in a changing climate. Coastal flood risk may be quantified as the probabilities of flood events and their potential consequences. Therefore, this thesis has addressed: (a) the statistical characterization of extreme events, and (b) the modeling of the associated consequences. The use of Monte Carlo methods is a common practice to undertake the statistical analysis, however, it can introduce a high computational burden for use in practical applications. Thus, in the first place, a methodology able to reduce this computational burden has been developed, which allows to end up in the economic quantification of risk after a robust exploration of all the possible sea conditions that might happen on the system. Next, in order to introduce climate variability on the risk analysis, different methodologies have been developed based on extreme value theory and statistical downscaling. The statistical downscaling technique relates a large scale predictor (sea level pressure fields) with a local predictand (multivariate sea states) by means of a clustering technique of synoptic atmospheric conditions, named weather types. This approach allows the statistical characterization of extreme events at different temporal scales (seasonality, interannual variability or climate change), by means of changes in the occurrence probability of the weather types. These statistical models might also help in the probabilistic design of coastal structures incorporating non-stationarity in the simulation of extreme nearshore sea conditions.