On the predictability of extreme events in records with linear and nonlinear long-range memory: Efficiency and noise robustness

► Short- and long-range analysis efficiency in predicting extreme events are compared. ► In linearly long-term correlated records, short-range analysis is always superior. ► In the presence of nonlinear memory, both approaches are of comparable efficiency. ► In the presence of intermediate additive white noise, long-range analysis is superior. ► Possible combinations of two methods do not lead to significant improvements.