Winter forecasting of September–October rainfall in the Murray–Darling Basin and south-eastern Australia

The atmosphere–ocean system governing Australian climate variability is fundamentally chaotic, characterised by high sensitivity to initial conditions. Major oceanic modes interact over large spatial and temporal scales. In such systems, predictability is not uniform in time but instead depends on the trajectory of the system through its underlying phase space. During austral winter, after the El Niño predictability barrier, sea surface temperature (SST) patterns in the Indian and Pacific Oceans mostly evolve slowly, providing a window of predictability for subsequent spring rainfall over south-eastern Australia. We focus specifically on winter prediction of rainfall during the September–October period, when Indian and Pacific Oceans exert a strong influence over south-eastern Australia. We formulate seasonal rainfall prediction as a reduced-order non-linear forecasting problem, embedding coupled Indian–Pacific Ocean variability into a low-dimensional state space and projecting it forward using deep neural networks. Variables include Niño 3.4, the Indian Ocean Dipole (IOD), the Indian Ocean meridional SST gradient and selected empirical orthogonal functions. Monthly time series of the variables then form the input into deep neural networks that project rainfall further into the future. Forecasts for the 2025 austral spring were generated during the winter and archived in the Mendeley database. Subsequent rainfall data demonstrated a high level of agreement with the forecasts, providing a validation of the method and supporting the hypothesis that chaotic yet conditionally predictable dynamics underpin spring rainfall variability in south-eastern Australia.