Author(s): A. D. Schepen; Q. J. Wang; D. E. Robertson
Linked Author(s):
Keywords: Climate Indices; Forecasting; Rainfall; Predictive Densities; Pseudo Bayes Factors
Abstract: In this study, we carried out an assessment of the ability of single climate index models to forecast rainfall in the next season across all of Australia on a 2.5 x 2.5 degree grid. The candidate predictors investigated were 13 climate indices that have been linked to Australian climate in the research literature. The models were assessed using the pseudo Bayes factor calculated from crossvalidation predictive densities. We examined the time-space evolution of the predictability associated with the El Nino Southern Oscillation (ENSO), Indian Ocean and Extratropical climate drivers. We found strong evidence for using climate indices to forecast rainfall in late winter and spring. In contrast, climate indices showed little ability for forecasting rainfall in autumn. We explain how rainfall forecasts are useful for streamflow forecasting.
Year: 2011