Empirical Orthogonal Function Analysis of Hawaiian Rainfall

Abstract

Empirical orthogonal function analysis was applied to monthly mean rainfall data at 63 stations in Hawaii encompassing a 37-year period. Major rainfall patterns in order to importance (E1?E3) proved to be trade wind, southwest wind and convective rainfall on an annual basis; trade wind, southwest wind and frontal rainfall during winter, spring and fall seasons; and trade wind, tropical disturbance and convective rainfall during summer. Trade wind rainfall (E1) explains most rainfall variance in summer and least variance in winter. Spectral analyses of the time-dependent coefficients for eigenvectors E1?E5 show annual, semi-annual, three-forths year, and 2?2½ year cycles. No spectral peaks relating to the 11- and 22-year sunspot cycles were found. Composite rainfall maps for wet and dry winter and summer half-years indicate the contributions that specific eigenvector patterns make to these anomalies. Comparisons between Hawaiian rainfall and E1 Niños reveal that most (not all) E1 Niño winters in Hawaii are dry. Lack of trade wind rainfall is the primary cause.