Trends in the southern oscillation phenomenon and Australian rainfall and changes in their relationship

Abstract

AbstractAn attempt has been made to investigate decadal‐scale trends in Australian rainfall and in the southern oscillation index (SOI) and their influence on the relationship between them. Monthly rainfall data from high‐quality stations in Australia (from 1900 to 1995), India, Sri Lanka and Tahiti are used. The relationship between the SOI and Australian rainfall is positive, but shows decadal‐scale variations during the past century. Although there were extended and severe El Niño events in the early 1990s and in 1997, Australia did not experience the expected severe rainfall deficiencies characteristic of previous events. However, severe drought conditions over eastern Australia were associated with a moderate El Niño event during 2002–03.Long‐term fluctuations of March–May (MAM) rainfall show high‐frequency variations, but trends during June–August (JJA), September–November (SON) and December–February (DJF) show low‐frequency or decadal‐scale variations. Trends and multi‐decadal fluctuations in all‐Australian spring (SON) and summer (DJF) rainfall are strongly dominated by rainfall trend fluctuations in northern and eastern Australia. Austral summer rainfall shows an increasing trend during the 1980s and 1990s, particularly in Queensland and New South Wales, despite the occurrence of extended and severe El Niños. However, some parts of New South Wales and Queensland experienced severe rainfall deficiencies during 2002–03 in conjunction with an El Niño event.The relationship between the SOI and rainfall on the interannual time scale is strong when the SOI and rainfall follow the same direction, but it is weak when they follow opposite directions on a decadal‐time scale. The poor correlation during the 1920s and 1930s was due to a slightly increasing trend in the SOI and a stronger decreasing trend in rainfall. A weakening in the relationship between the SOI and rainfall in recent years, after the mid‐1970s, is due to a small increase in rainfall in the 1980s and 1990s and a strong decrease in the SOI. Rainfall trends were enhanced (stronger decreases or increases) when the influence of the SOI (or El Niño‐southern oscillation (ENSO)) was removed. Enhanced increases and decreases are particularly strong during SON and DJF, when the ENSO phenomenon is at the mature stage and also the influence on Australian rainfall is strong. The increasing trend in rainfall during the 1980s and 1990s in some parts of eastern Australia and the decreasing trend in the SOI result in more rainfall for a given SOI compared with the same SOI during the previous period, i.e. before the mid‐1970s. A similar analysis was carried out for two periods, before and after 1972, for Tahiti, India and Sri Lanka. The upward or downward shift in regression lines is very clear during the season, that shows a strong relationship between rainfall and the SOI. Moreover, strengthening or weakening of the relationship between rainfall and the SOI is largely dependent on their multi‐decadal variations and trends during the past century.Increases in rainfall during the 1980s and the 1990s and decreases in the SOI have weakened their relationship, both in Australia and India. Such a relationship gives more rainfall for a given SOI after 1973. The pattern was reversed for Sri Lanka, where rainfall during the second intermonsoon season has decreased. Analyses of trends in temperature at Darwin and Tahiti and of rainfall over Australia, India, Tahiti and Sri Lanka suggest a regional‐scale change in climate, whereas the SOI reflects a change in the large‐scale circulation pattern over the Indo‐Pacific region after the mid‐1970s. Copyright © 2004 Royal Meteorological Society