Long‐range prediction of hawaiian winter rainfall using canonical correlation analysis

Hawaiian rainfall is teleconnected to short‐term climate variability in the Pacific Ocean. The summer Southern Oscillation Index (SOI) and summer sea‐level pressure (SLP) over the North Pacific are used as predictors, and the following winter rainfall indices from three islands of Hawaii are used as predictands. To consolidate the large data array of the SLP field prior to prediction experiments, lagged correlation and empirical orthogonal function analyses are used. Canonical correlation analysis has been used for predicting Hawaiian winter rainfall. Among many schemes tested, predictions. Cross‐validation techniques have also been used to estimate the “overall” forecast skill of various schemes, and the results are consistent with those from prediction experiments. Winter rainfall in Hawaii can be predicted with a good degree of success two seasons in advance using the summer SOI and the first four eigenmodes of summer SLP over the North Pacific as predictors.