Developing and validating a forecast/hindcast system for the Mediterranean Sea

In this work a study of the performances of the WAVEWATCH III (WWIII) model in the Mediterranean basin is presented. Analysis is carried out referring to seventeen case studies corresponding to heavy storms in northern Tyrrhenian basin (Ligurian sea) in the last twenty five years. Simulation results are validated using buoy data provided by the Rete Ondametrica Nazionale (RON). An analysis of the usage and performances of different statistical error indicators is provided, showing that widespread NRMSE indicator is biased towards models that underestimate prediction. The well established source terms parameterization by Tolman and Chalikov (1996) has been compared with the one proposed by Ardhuin et al.( 2010), set up with the parameterization by Bidlot et al., (2005), and with the set known as ACC350. The obtained results reveal that the ACC350 parameterization works better in severe conditions, though tends to overestimate wave height and underestimate period. A further sensitivity analysis in the parameters space is carried out around ACC350 parameterization, finding that variations in the default set of parameters, involving small wave growth reduction or an increase of energy dissipation, lead to a slight improvement of the overall simulation performances. © Coastal Education & Research Foundation 2013.