Seasonal to interannual rainfall probabilistic forecasts for improved water supply management: Part 2 - Predictor identification of quarterly rainfall using ocean-atmosphere information

This paper is the second in a series of three in the current issue that pre sent a framework for long-term rainfall probabilistic forecasts. The first paper of the series presented the partial mutual information (PMI) criterio n as an efficient basis for system predictor identification in the probabil istic forecast context. This paper presents applications of the PMI criteri on to identify the best predictors of quarterly rainfall at Warragamba dam near Sydney Australia, using a range of ocean-atmosphere predictor variable s. Two separate prediction scenarios are considered. The first scenario involv es the use of three commonly used El Nino Southern Oscillation indices as p redictors of the Warragamba dam quarterly rainfall. The second scenario use s sea surface temperature anomalies averaged over 5 degrees latitude by 5 d egrees longitude blocks as the plausible system predictors. A reconstructed sea surface temperature anomaly dataset extended to 1856 is used in this a nalysis. The usefulness of the predictors from both scenarios is evaluated by forecasting rainfall for seasonal to interannual lead times, using a Gen eralised Additive Model. Details of the rainfall probabilistic forecasts us ing the predictors identified under the two scenarios are presented in the last paper of this three-paper series. (C) 2000 Elsevier Science B.V. All r ights reserved.