SCIENCE OF SEASONAL CLIMATE PREDICTION

Future directions in the science of seasonal prediction
by Oscar Alves, BMRC

Dr Oscar Alves has a PhD in ocean modelling from University of Reading, UK, and he spent seven years at UK Met Office working on ocean modelling and four years at ECMWF working on the first ECMWF dynamical seasonal prediction system. He joined BMRC in 2000 as leader of the dynamical seasonal prediction project (POAMA). Since 2005 he has headed the Ocean and Marine Forecasting Group, which includes POAMA and a range of other ocean and climate related projects. His research interests include dynamical seasonal prediction, ocean modelling, data assimilation, and tropical climate variability.

The socio-economic benefit from accurate seasonal forecasts now justifies the support for a range of activities from basic forecast model development to applications studies. Coupled model seasonal forecasts are now becoming competitive with more traditional statistical based forecasts. Coupled models offer the long term potential to provide seasonal predictions in an era of possible climate change, whereas, by definition, statistical approaches cannot be used if climate relationships are changing. However, many problems still exist with coupled models, most suffer significant climate drift and most have trouble adequately simulating some of the details of some fundamental modes of variability. This leaves a lot of room for improvement.

Coupled model forecast systems use the latest ocean and atmosphere observations for the initialisation of coupled forecasts. Over the last few years there has been significant improvements in ocean observing systems and these improvements are important for more skilful real-time forecasts. However, it is difficult to assess the skill of the real-time forecasts using retrospective hind-casts over say the last 25 years, because during the 1980’s and 1990’s the ocean observing network was relatively sparse compared with the last five years.

POAMA is the current operational dynamical seasonal prediction system at the Bureau of Meteorology. The system was developed jointly with CSIRO and Land and Water Australia. A new version, POAMA-2, has been built and will go operational early 2007. POAMA-2 has several enhancements and for the first time will provide rainfall forecasts in addition to El Niño forecast. In the longer term, versions of POAMA will use the ACCESS coupled model as one of its main components. The ACCESS model is the new earth system model being developed jointly by BMRC and CSIRO for a range including numerical weather prediction, seasonal prediction and climate change.

This talk reviews the current trends and issues in seasonal prediction and discusses the future directions of the science.