Observed and simulated relationships amongst ENSO, the IPO, and rainfall variability in eastern Australia

Abstract

Understanding changes in past and future rainfall variability can improve societal, environmental and economic decision-making. Global climate models (GCMs) are commonly used to improve the understanding of rainfall variability; however, accurately simulating the variability of precipitation is difficult as many processes with different spatial and temporal scales influence precipitation. Interactions between climate drivers, which influence precipitation over wide temporal and spatial scales, often have the greatest impact on rainfall variability. Currently, the ability of GCMs to simulate interactions between climate drivers and their influence on rainfall is not well known. This thesis uses correlation and composite analysis to investigate how the joint interaction between two climate drivers, the El Niño Southern Oscillation (ENSO) and the Interdecadal Pacific Oscillation (IPO), influences eastern Australian rainfall variability, and whether or not this relationship is simulated by a GCM optimised to simulate the Australian climate. To do this, the Australian Water Availability Project (AWAP, run 26j) gridded observational dataset is compared with the Australian Community Climate and Earth-System Simulator (ACCESS1.3) GCM for the years 1900–2005. The ENSO–rainfall teleconnection is stronger and geographically broader than the IPO---rainfall teleconnection, and ENSO tends to influence lower-latitude rainfall whilst the IPO tends to influence higher-latitude rainfall. The ENSO–rainfall relationship is strongest in austral spring and weakest in austral autumn. The IPO exhibits a strong negative correlation with rainfall in austral summer, but influences rainfall the most in austral autumn, when the ENSO signal weakens. ACCESS1.3 simulates the ENSO teleconnection with rainfall with reasonable accuracy, although its simulation of the seasonality and variability in regional rainfall requires improvement. The model appears to represent the IPO as an ENSO-like phenomenon, and does not simulate the spatial or temporal features that characterise the IPO. The ENSO–IPO interaction exhibits its strongest influence on eastern Australian rainfall in austral summer and autumn. The influence of the joint ENSO–IPO interaction on rainfall reflects the interplay between the inter-annual and inter-decadal scales of the climate drivers. Stratification of rainfall into the nine combinations of the positive, neutral and negative states of the ENSO and IPO reveals that ENSO has a strong influence on rainfall variability, but that the IPO modifies the relationship between ENSO and rainfall. The IPO negative phase enhances the ENSO--rainfall relationship; the IPO positive phase attenuates this relationship; and the IPO neutral phase results in a slight decrease in rainfall. In the case of moderately extreme ENSO and IPO events, these relationships change somewhat. Rainfall variability is influenced by ENSO, but is modified by the IPO neutral and positive phases, and in the case of the most extreme ENSO and IPO events, only the IPO appears to influence rainfall variability. ACCESS1.3 does not simulate these relationships. If the accuracy of modelled precipitation is to increase, the physical processes through which climate drivers interact should be incorporated into GCMs. Show more