Conditional nonlinear optimal perturbation and its applications

Conditional nonlinear optimal perturbation (CNOP) is proposed to study the predictability of numerical weather and climate prediction. A simple coupled ocean-atmosphere model for ENSO is adopted as an example to show its applicability. In the case of climatological mean state being the basic state, it is shown that CNOP tends to evolve into El Ni (n) over tildeo or La Ni (n) over tildea event more probably than linear singular vector (LSV) on the condition that CNOP and LSV are of the same magnitude of norm. CNOP is also employed to study the prediction error of El Ni (n) over tildeo and La Ni (n) over tildea events. Comparisons between CNOP and LSV demonstrate that CNOP is more applicable in studying the predictability of the models governing the nonlinear motions of oceans and atmospheres.