The Southern Oscillation Index as a predictor of seasonal rainfall in the arable areas of the inland Australian subtropics

Abstract

A detailed study has been carried out in four regions in the subtropics of Eastern Australia to determine the relationship between the Southern Oscillation Index (SOI) and subsequent seasonal rainfall. The period studied was from 1915 to 1991 for 3-monthly periods of spring (SON), summer (DJF), autumn (MAM) and winter (JJA). The 3-monthly prior SOI values were plotted against seasonal rainfall of the four regions and four seasons. These data were widely scattered but with a linear trend showing increased seasonal rainfall as the SOI increased. Linear trends were plotted for each season and region. Comparisons were made between the use of the ACE algorithm, which transforms the SOI and rainfall data, and the use of linear trends. Polynomials were used to calculate equations for each region and season, but only spring and summer produced satisfactory ACE functions. Estimates were made of spring and summer rainfall relative to prior SOI values for each region. While the SOI as a predictor of rainfall broadly estimates spring and summer rainfall, this variable has limited usefulness on its own. One of the options available with the ACE program is that additional independent variables can be added as required. Current research suggests that sea surface temperature data from specific ocean areas surrounding the Australian continent is the most useful additional variable at present. However the complexity of such an analysis is greatly increased.