An investigation of the nature of variability in Australian rainfall using wavelets

Abstract

The method of wavelets was used to assess the influence of medium- to long-term climate variability on three aspects of Australian rainfall time series: (1) the total annual rainfall, (2) the total annual number of wet days, and (3) the maximum annual daily rainfall. The influence of climate variability could be readily established on the first two of these series, although not with the third series. This may be attributed to the greater level of random behaviour in the intensity of extreme events, which make the influence of climate variability more difficult to quantify. Furthermore, based on the results from the total annual rainfall analysis, three regions with distinct spectral signatures were identified. Significant differences between the spectral signatures for each region and the spectral signature of the Nino 3.4 time series were shown, even for cases where the rainfall time series was highly correlated with Nino 3.4 or for cases where there was significant variability in the 3 to 6 year band that is commonly associated with ENSO. It is therefore clear that the method of wavelets is able to show significant additional information on the nature of Australian rainfall time series than that provided by correlation analysis or traditional spectral techniques such as Fourier analysis alone.