We chose to use a percentile-type parameter to avoid the effects of negative HsHs. Regarding the extreme wave climate, we analyze the 50-year return value of HsHs, which was computed as in Casas-Prat and Sierra (2013) using a Generalized Pareto Distribution model. Fig. 16 and Fig. 17 show the median HsHs projected using Setting 5, with the predictors being derived before and after applying B-Raf inhibitor drug the adjustments to the model data, respectively. The upper panels show the present-day climatological values; whereas the

lower panels show the projected changes in future climates that are expressed as a portion of the present-day climatological value. Each column corresponds to one of the five sets of model simulations (see Section 3.2). As shown in Fig. 16, HIR_E model has a clear positive bias (overestimation of projected HsHs). The other models show more similar present-day wave climates, which have much smaller positive

biases. When forced by the same GCM ECHAM5, all four RCMs (HIR_E, RAC_E, REM_E, RCA_E) project future changes that share a common tendency for HsHs to increase in the NE part of the domain (up to 10%). An increase is projected for the area near the Gulf of Genoa, suggesting Navitoclax mw an increase in future cyclone activity in this important cyclogenesis area in the Mediterranean (see Section 2.1). This is consistent with the Urease projected increase in mean gust of gust event days in winter (October–March) reported by Schwierz et al. (2010), who analyzed CHRM (a RCM) simulations with the ECHAM5 and HadCM3/HadAM3 lateral boundary conditions. In the SW part of the domain, HsHs tends to decrease (up to 10%) but the extent of decrease varies between RCMs. HIR_E projects a more pronounced decrease; whereas the REM_E and RCA_E models project much more limited decreases. Similar patterns of projected mean wave climate were obtained by Casas-Prat and Sierra (2013) using dynamical downscaling. However, they simulated the area of increase (in the

NE part of the domain) closer to the Catalan coast. On the other hand, RCA_H (which is forced by the HadCM3Q3 global model) projects a general decrease of HsHs (up to 10%) over the entire domain (especially in the SE part). Close to the east-facing coasts, HsHs reduction is smaller and in some stretches it tends to remain the same or even to slightly increase. This spatial pattern of change is in agreement with what is projected by global models as presented in the study of Donat et al. (2010) and by the regional dynamical downscaling of Casas-Prat and Sierra (2013). Donat et al. (2010) found an increase of E flow for a model similar to HadCM3Q3 but a tendency of the increased W flow for those forced by the ECHAM5 global model.