**3. ICCAI: seasonal climate prediction for the Pacific Island nations**

CREWS-PNG is building on the success of earlier Pacific projects such as the Pacific Adaptation Strategy Assistance Program (PASAP) and the Pacific-Australia Climate Change Science Adaptation Planning (PACCSAP) program undertaken under the Australian's Government International Climate Change Adaptation Initiative. Under those Pacific projects, a number of regional consolidated databases and webbased information tools were developed, to assist 15 partner countries in the Pacific with decision-making in climate change adaptation and disaster risk reduction.

It has been demonstrated that seasonal climate prediction tools are a valuable means to assist with climate change adaptation and disaster risk reduction [18]. For decades, seasonal climate prediction in Australia and the Pacific Island Countries was based on statistical models [10]. However, in a rapidly changing environment, the skill of statistical models which are solely based on past records is expected to deteriorate. Hence dynamical climate models, which are based on laws of physics rather than past relationships, are preferable for forecasting climate on time scales from weeks to months and beyond.

It has been shown that using the outputs of the dynamical climate model Predictive Ocean and Atmosphere Model for Australia (POAMA), skilful predictions of regional rainfall could be provided 2–3 months in advance [19, 20]. POAMA was developed [21] and used by the Bureau of Meteorology operationally to produce its seasonal climate outlooks from 2013. It is a state-of-the-art dynamical (physicsbased) forecast model which used ocean, atmosphere, ice and land observations to initiate the model and produce outlooks for the season ahead.

Recently, a new higher resolution model—the Australian Community Climate and Earth-System Simulator (ACCESS)—has succeeded POAMA to further

#### *Drought - Detection and Solutions*

enhance the Bureau's capability in weather and climate prediction. Since August 2018 the Bureau's operational climate forecast system for multi-week, monthly, seasonal and longer-range climate outlooks is ACCESS-Seasonal (ACCESS-S). The atmosphere and land model components of ACCESS-S operate at an approximate global resolution of 60 km, providing far greater detail than POAMA which had an approximate resolution of 250 km. The multi-week and seasonal performance of ACCESS-S has been evaluated for Australia based on a 23-year hindcast set and compared to the previous operational system, POAMA [22]. However, hindcast skill and calibration have not been assessed for other countries in the region including PNG. One of the key tasks for CREWS-PNG is to assess and utilize outputs of ACCESS-S in order to produce accurate multi-week and seasonal climate outlooks that will improve drought prediction for the country.
