Most state-of-the-art general circulation models (GCMs) are unable to predict well the year-to-year variation in the western North Pacific summer monsoon (WNPSM). Good skill is unrealized as model biases distort the distribution of the predicted monsoon rainfall in individual models, despite the fact that the wind fields are relatively well reproduced. However, a multi-model ensemble (MME) approach can considerably reduce biases of rainfall simulations. Here we demonstrate that, by using combinations of hindcast runs from thirteen different models participating in the APEC Climate Center (APCC) MME forecasts, the WNPSM rainfall can be skillfully predicted. It is also found that the coupled MME (CMME) mean predictions which comprise Tier-1 prediction systems outperform each individual model in capturing both the spatial and temporal variability of WNPSM circulation.

1. INTRODUCTION

2. METHODOLOGY

2.1 Data

2.2 Methods

3. RESULTS

3.1 JJA mean fields of 850 hPa wind and precipitation in the WNP-EA region

3.2 WNPSM in observations, GCMs and MME onto the observed WNPMI

3.3 Interannual variability of the WNPMI in observations, GCMs and MME

3.4 Model combinations of simple composite MMEs based on the leading WNPSM mode

3.5 Leading mode of the WNPSM interannual variability in simple composite MMEs

3.6 Predictability of the WNPSM interannual variability in simple composite MMEs

4. CONCLUDING REMARKS