Improving forecast accuracy with improved Indian Ocean initialisation

Research objective

Improve predictions of conditions in the Indian Ocean and, ultimately, predictions of regional climate for western, southern and eastern Australia.

Project duration

2010 – 2013


POAMA uses sophisticated techniques to ingest oceanic observations which it then uses to initialise model forecasts and estimate the ocean state at each point in time.

We included new ocean observations (sea surface temperature and altimeter) and accounted for ocean–atmosphere coupling.

We compared the skill of Indian Ocean sea surface temperature predictions from POAMA-2P and POAMA-1.5 with those from other international models.

During the second half of the year, when the Indian Ocean Dipole (IOD) develops, the forecast skill from the POAMA model is as good as the best international models and that model error is a common problem for limiting forecast skill of the IOD.

We trialled several initialisation improvements with POAMA: the initialisation of ocean salinity played an important role in the improvements seen in POAMA-2. This is believed to be the way forward in initialisation and our research is at the leading edge. A second major enhancement trialled the direct assimilation of sea surface temperature observations, which means that they will now impact the ocean sub-surface. Including altimeter data did not improve IOD predictions.

This project enabled POAMA scientists to take part in an internation model comparison and benchmarking.

Real-time analysis products are now available on the POAMA website, allowing for real-time monitoring of ocean temperatures, salinity, currents and sea level. It also provides information about the ocean anomalies that are being used to initialise the POAMA forecasts.

Read the final report as MSOffice Word [31 kb] or PDF [105 kb].

Research contact

Dr Oscar Alves

Centre for Australian Weather and Climate Research

Phone: 03 9669 4835

You may also like...