Triple annual maximum series in wave climate analyses

Abstract

A recurring problem in wave climate analyses is the need to predict long term events from short duration records. The relative sparsity of the observational record is enhanced by judicious recognition of near-maximum events, in addition to the annual maximum events. This paper pursues the application of triple annual maximum series. Given any extreme value distribution, the theoretical distributions for the annual second largest monthly wave and the annual third largest monthly wave are established. A maximum likelihood method is proposed to fit these simultaneous distributions to the triple annual maximum series. An application of the method to the historical record at the Farallon Is. off San Francisco adopts several of the more common extreme value distributions and demonstrates the potential of triple annual maximum series in enhancing the reliability of distribution fits. Nevertheless, the common practice of extrapolating short duration records to long term events remains precarious.