Statistical behaviour of Adelaide's rainfall - Is climate change detectable?

The South Australian climate is changing towards less rainfall and consequently water security problems will be intensified (IPCC and CSIRO, 2007). In this study we attempted to identify the statistical behaviour of point rainfall measured at the Adelaide airport. Statistical moments, lag1 autocorrelation, the Buishand's Q test for homogeneity, the Mann-Kendall test for trend and wavelet analyses for shift and temporal variability were carried out at fine (< 12 hour) and at coarse (monthly) temporal resolutions. Series of rainfall intensities at finer temporal scales and rainfall depths and proportion dry ratios at monthly scales were accumulated from the high resolution rainfall dataset. Homogeneity of rainfall intensity was found to increase as the temporal scale increases. Both rainfall intensities and monthly rainfall depths were found to be serially correlated. Except for July, the 2^nd, 3^rd and 4^th statistical moments of rainfall intensities decrease as the scale increases. While no statistically significant trends were found at finer and monthly scales using the Mann-Kendall test, there were indications that trends are more likely as the temporal scale increases. Wavelet power spectra also showed this scenario noticeably. At very fine temporal scales (< 1 hour) rainfall intensities have large temporal variability and no trends or shifts in frequency level were detected. At the 12 hour and monthly temporal resolutions, statistically significant changes of dominant frequency level were observed in some months. Interestingly these changes were observed to happen after 1990.