Abstract Author: Guomin Wang
Abstract Title: Seasonal Prediction Research and Development at the Australian Bureau of Meteorology
Abstract: The dynamical model ENSO forecast skill has been dramatically improved in the past two decades due to continuing effort in observing and understanding the coupled ocean-atmosphere system. At the Australian Bureau of Meteorology (BoM) there is an ongoing call for exploring dynamical model based seasonal prediction beyond ENSO, most noticeably drought prediction. In this talk I will report on our latest seasonal forecast system POAMA (Predictive Ocean Atmosphere Model for Australia) development work, and several research and application studies.
Australia experienced near normal rainfall in spring 1997, when a strong El Nino was developing. In contrast severe drought ravaged the country in 2002, when a moderate El Nino was evolving. POAMA successfully forecasted rainfall anomalies over Australia for the two years. Diagnostic analyses indicate the success may be related to POAMA’s capability to predict subtle SST difference between the two El Niños.
The Leeuwin Current (LC) flows poleward along the west coast of Australia. Interannual variations of the LC exhibit strong connection with ENSO and have profound impacts on regional climate. Direct seasonal prediction of LC with present models is not feasible due to insufficient resolution. Using forecasts of the heat content along northwest shelf of Australia (hereafter HCNW) and the SST Niño3.4, both of which show higher skill from POAMA, and strong observed correlation with Fremantle Sea Level anomalies (FSLA), a downscaling model, which transfers skill in prediction of HCNW and Nino3.4 SST to FSLA, has been developed. The skill measured by anomaly correlation coefficient (ACC) between the observed and the forecast FSLA is above 0.6 to nine month lead time from the downscaling scheme, whereas ACC from an empirical scheme is much lower.