A comprehensive assessment of CFS seasonal forecast skills over the tropics

In the present research, the 15-member ensemble retrospective forecasts (hindcasts) performed with the NCEP Climate Forecast System (CFS) for the period 1982-2005, as well as real-time forecasts for the period 2006-09 are assessed for seasonal prediction skills over the tropics, both from deterministic (anomaly correlation) and probabilistic (Heidke skill score and rank probability skill score) perspectives. The probabilistic scores convey the errors associated with the model's seasonal prediction skill. In addition, diagnostics such as persistence, root-mean square error and composite analysis are performed.

Based on both deterministic and probabilistic measures, CFS demonstrates high skill (even at 6-month lead) in predicting tropical central to eastern Pacific sea surface temperature (SST) anomalies, including the different flavours of El Nino. The model is also skilful in forecasting the associated precipitation anomalies over the equatorial Pacific as well as capturing the teleconnection to the tropical Indian Ocean. Owing to the overall success in forecasting the tropical Indo-Pacific climate anomalies, we examined the seasonal skill of precipitation over three regions encompassing the Pacific Islands. The salient merits and limitations of the model in forecasting the regional rainfall anomalies, in conjunction with the model's skill of the real-time forecasts will be discussed.