HR: 17:15h
AN: H24B-06    [Abstracts]
TI: An exploratory study of seasonal rainfall variability in Australia using Independent Component Analysis
AU: Westra, S
EM: seth@civeng.unsw.edu.au
AF: School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW 2052 Australia
AU: * Sharma, A
EM: a.sharma@unsw.edu.au
AF: School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW 2052 Australia
AB: Component extraction techniques have been used frequently by climate and water resources researchers to analyse high dimensional data sets. In this study we explore the application of a relatively new technique known as independent component analysis (ICA) to time series of Australian rainfall, as an alternative to the better known principal components analysis (PCA). The primary distinction between these two techniques is that while PCA uses only second order statistics to obtain uncorrelated components, ICA attempts to maximise independence between the components through the use of higher order statistics. Using a synthetic study that is designed to highlight important characteristics of Australian rainfall time series, we show that while PCA is better suited to the tasks of variance maximisation and dimension reduction, ICA is fundamentally more suited to ensuring the statistical independence of the extracted components and in certain cases is also capable of determining the underlying causes of this variability. As a result, we consider the ICA methodology to be more suitable for formulating a statistical basis for predicting rainfall, since the independence criterion allows for the formulation of distinct statistical models for each identified component. The sensitivity of the ICA to several factors including record length, the number of components extracted, the amount of noise in the data and the need for dimension reduction before the ICA, are examined in detail. The ICA technique is then applied to seasonal rainfall time series from over 200 rainfall gauges located around the Australian continent. The physical interpretability of the extracted ICs is assessed based on existing knowledge on the underlying causes of rainfall variability in Australia, and conclusions on the suitability of the technique for statistical forecasting are drawn.
DE: 1812 Drought
DE: 1833 Hydroclimatology
DE: 1854 Precipitation (3354)
SC: Hydrology [H]
MN: 2004 AGU Fall Meeting