Long range daily ocean forecasts in the Indo-Pacific oceans with ACCESS-S1

Currently most ocean forecast systems provide daily eddy resolving predictions 7 days or so in advance. We first investigate the potential for longer range daily ocean forecast up to 28 days using the coupled system ACCESS-S1 in which the ocean component is eddy permitting. The daily ocean forecasts over the hindcast period 1990–2012 in the Indo-Pacific oceans have been evaluated, using persistence skill as a reference. Our results show that the sea surface temperature (SST) biases are in the range of ±0.3 °C in most regions within the first 28 days of the forecast. The model’s predicted correlation skill of SST anomaly (SSTA) is higher than persistence skill in most areas beyond the first few days up until the 28-day lead time. The skill of SSTA indices averaged in different domains suggest that the model produces more skilful predictions in the North Indian Ocean, South Indian Ocean, western tropical Pacific and Leeuwin Current regions than the corresponding persistence forecasts. The model’s SSTA prediction skill in the eastern tropical Pacific is comparable to the persistence skill which is very high. The model’s correlation skill for sea surface height anomaly (SSHA) can beat persistence skill beyond the first week and the model’s skill has a great advantage with a longer lead time. The daily prediction of the larger mesoscale eddies that shed off the East Australia Current is also investigated here. The spatial correlation between the model predicted and observed SSHA is applied to measure the model prediction skill for these eddies. Our results suggest that the averaged correlation for all forecasting cases at the same lead time is above 0.5 up to 14-day lead. It indicates that ACCESS-S1 has some skill in predicting these eddies, since the strength and track of the eddies in the Tasman Sea are highly complicated.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.