“…In the present paper, for each RCP scenario, two different ensemble models E1 and E2 are proposed; the former only considering the set of significant MS models, the latter including all the M models in Table . The ensemble estimation is built by performing a weighted average of quantiles for a given return period T ; this was preferred to an equally weighted average to account for the different performance of climate models in reproducing historical observations, considering poorly‐performant hindcasts to be “less reliable” (Bhat, Haran, Terando, & Keller, ; Reder et al, ; Rianna, Reder, Mercogliano, & Pagano, ): where w j is the weight associated to each model j , that should be selected as a metric representative of the “distance” between the observed and the current sample. Moreover, in Equation for .…”