The spring prediction barrier of ENSO in retrospective prediction experiments as shown by the four coupled ocean atmosphere models

Based on the dynamic and statistical analysis methods, this study analyzes recent 22-year (1982-2003) retrospective ENSO prediction performed by the three coupled GCMs that participate in the Climate Prediction and its Application to Society (CliPAS) project and by a coupled GCM namely FGOALS g developed at LASG/IAP. The seasonal dependence of prediction error growth for both the growing and decaying phases of El Nino/La Nina events is presented. All the four coupled models show considerable ENSO prediction skill, and the so called “Spring Prediction Barrier” (SPB) is also very evident for the each retrospective prediction experiment. The further analysis suggests that SPB is strongly associated with the prediction error growth during the spring, in particular, the growth rate of prediction error is the strongest during the spring for El Nino events and the growing phase of La Nino a evens, but it does not depend on the season for the decaying phase of La Nina events. We have also found significant asymmetry in the growth rate of prediction error of SST anomaly between El Nino and La Nino a events. By analyzing the regression, we found that the air-sea interaction is the most unstable in the spring, which favors rapid growth of prediction error in this season and then results in SPB in the retrospective ENSO prediction experiments.