Propagation of uncertainty across modeling chains to evaluate hydraulic vulnerability in coastal areas

The aim of the thesis is to investigate the propagation of the uncertainties from meteorological to coastal forecasts, in order to obtain a better understanding of the uncertainties associated to the numerical modeling systems. The first phases focused on the parameter settings of the morphological model XBeach, as source of uncertainties within the model itself. This was done by means of a sensitivity analysis of the model that allowed to characterize how the model responds to changes in input, with an emphasis on finding the input parameters to which outputs are the most sensitive. Moreover, an estimate of how the uncertainties propagate within the numerical modeling chain was made by means of the ensemble technique. Moving from a single-deterministic to probabilistic forecasts, it is possible to give some useful indication of the forecast reliability. Therefore, the meteorological Limited Area Ensemble Prediction System COSMO-LEPS was used to generate 16 different meteorological forecasts that were used to force the wave\oceanographic models SWAN and ROMS and finally the morphological model XBeach. The study focused on two different storm events both occurred in the autumn 2015-winter 2016 on the Emilia-Romagna coasts.The results showed that, in both cases, the uncertainties of the wind and pressure fields clearly propagated through to the oceanographic models up to influence the coastal forecasts. The accuracy of the forecasts of the oceanographic and morphological models is largely dependent on the quality in wind data. However, extension of the ensemble approach to the coastal areas showed encouraging results and suggested, as a future development, the possible optimization of the system by using a meteorological ensemble built in such a way as to optimize the spread in terms of the surface variables used to drive the marine-coastal model components.