Improving Prediction of Two ENSO Types Using a Multi-Model Ensemble based on Stepwise Pattern Projection Model

Abstract:

This study focuses on improving the prediction skill of two types of ENSO using a Multi-Model Ensemble (MME) method based on 5 dynamical models datasets from the North American Multi-Model Ensemble (NMME) project for the period 1982-2010. The strength of the MME lies in a statistical error correction procedure using a stepwise pattern projection model for all contributing models. The prediction skill of the proposed MME show an improvement over most tropical Pacific regions. With regard to the two types of ENSO, improvements in predictive skills of the proposed MME are particularly evident for the Niño indices for most lead months. The differences between Eastern Pacific (EP) and Central Pacific (CP) events are more pronounced in corrected forecasts compared with uncorrected ones. In addition, we find that though the corrected MME is not the best one among corrected forecasts, the central positions of the corrected MME is closer to the observed center than that of the uncorrected MME. The results indicate that removing systematic biases of each model member by using a good error correction method before applying MME method can provide an effective way of empirically improving ENSO forecasts, particularly with regard to forecasts of the two ENSO types.