A Study of Predictable Patterns for Seasonal Forecasting of New Zealand Rainfall

Abstract

A recently developed variance decomposition approach is applied to study the causes of the predictability of New Zealand seasonal mean rainfall. In terms of predictability, the Southern Oscillation is identified as being the most important cause of variability for both the winter and summer New Zealand rainfall, especially for the North Island. Indian Ocean sea surface temperature variability and the Southern Hemisphere annular mode are the second most important causes of variability for winter and summer rainfall, respectively. Based on this study, a statistical prediction scheme has been developed. May Niño-3 (5°N?5°S, 150°?90°W) SSTs and March?May (MAM) central Indian Ocean SSTs are identified as being the most important predictors for the winter rainfall, while September?November (SON) Niño-3 SSTs, November local New Zealand SSTs, and the SON Southern Hemisphere annular mode index are the most important predictors for the summer rainfall. The predictive skill, in term of the percentage explained variance for the verification period (1993?2000) is nearly 20%, which is considerably higher than that achieved previously.