A kind of initial errors related to spring predictability barrier'' for El Nino events in Zebiak-Cane model

[1] Seasonal dependence of initial error growth for El Nino-Southern Oscillation (ENSO) in Zebiak-Cane model is investigated by using a new approach, i.e. conditional nonlinear optimal perturbation (CNOP). It is found that CNOP-type error tends to have a significant season-dependent evolution, and produces most considerable negative effects on the forecast results. Therefore, CNOPs are closely related to spring predictability barrier (SPB). On the other hand, some other kinds of initial errors, whose patterns are different from those of CNOPs, have also been found. Although the magnitudes of such initial errors are the same as those of CNOPs in terms of the chosen norm, they either show less prominent season-dependent evolutions, or have trivial effect on the forecast results, and consequently do not yield SPB for El Nino events. The results of this investigation suggest that the CNOP-type errors can be considered as one of candidate errors that cause the SPB. If data assimilation or (and) targeting observation approaches possess the function of filtering the CNOP-type or (and) other similar errors, it is hopeful to improve the prediction skill of ENSO.