On the occurrence and importance of Paretian-tailed distributions in hydrology

Type-I and type-II extreme-value distributions are used to compare the tail behaviour of empirical flood data. The frequency with which type-II provides a better fit suggests that Paretian- or heavy-tailed distributions occur very often. The impact of tail behaviour on the prediction of extreme events is investigated, and a heuristic method for getting bounds on extreme-event magnitude is suggested.