Predictability and prediction skill of the boreal summer intra-seasonal oscillation in BCC_CSM model

Abstract

The boreal summer intra-seasonal oscillation (BSISO) is the predominant sub-seasonal variability over the East Asia (EA) and western North Pacific (WNP) region and critical for seasonal forecast of the EA summer monsoon. This study examines the theoretically estimated predictability and practical prediction skill of the EAWNP BSISO in the Beijing Climate Center Climate System Model version 2 (BCC_CSM2.0), which is one of the participants in the Sub-seasonal to Seasonal Prediction Project. The results from the uninitialized free run of BCC_ CSM2.0 show that the model reasonably simulates EAWNP BSISO in terms of its variance, propagation, and structure. Measured by the bivariate correlation (> 0.5) and root mean square error (< 2) between the predicted and observed real-time BSISO index, the prediction skill and predictability of EAWNP BSISO are about 14 and 24 – 28 days, respectively. The initial/target strong BSISO cases have a relatively higher prediction skill than the initial/target weak BSISO cases. For the theoretically estimated BSISO predictability, a similar dependence on target amplitude occurs in the model, while no significant dependency on initial amplitude is found. Moreover, diagnosis of the phase dependence reveals that BSISO is less skillful for the prediction starting from active or active-to-break transition phases of WNP rainfall, whereas it is more predictable when prediction is targeting extreme dry/wet phases of WNP rainfall. Finally, systematic errors are found in BCC_CSM2.0 such as the under-estimation of BSISO amplitude and the faster phase speed.