POAMA (Predictive Ocean Atmosphere Model for Australia) is a coupled ocean/atmosphere model developed jointly by the Bureau of Meteorology Research Centre (BMRC), Melbourne and CSIRO Marine Research (CMR), Hobart. It is based on the latest version of BMRC's unified climate/weather atmosphere model (BAM) and the Australian Community Ocean Model (ACOM2). The ocean model is initialised with oceanic fields produced by an ocean data assimilation scheme developed at BMRC. It uses optimum interpolation to combine ocean sub-surface observations with a model background field. The atmosphere model is initialised with fields from operational weather analyses.

The POAMA system is run in real-time by the operational section of the Bureau of Meteorology. Every day an eight-month forecast is produced. This uses initial conditions from the data assimilation systems, which is also run in real-time using observations from the Global Telecommunications System. This means that each forecast is initialised with the very latest oceanic and atmospheric states.

The operational system and latest results will be described. The initial focus of POAMA is the prediction of El Nino. Results will be presented based on hind-casts produced for the 1980s and 1990s. These show that the skill of POAMA forecasts is at least as good as other international models. Several scientific issues related to dynamical seasonal prediction will also be discussed, in particular: the role of intra-seasonal variability in El Nino forecasting, the model's ability to represent intra-seasonal variability and the importance of atmospheric initial conditions.