ISSN 1002-1027  CN 11-2952/G2

Acta scientiarum naturalium Universitatis Pekinensis

Previous Articles     Next Articles

Nonlinear Prediction of ENSO

LI Kunyu, LI Xiaodong   

  • Received:2006-03-27 Online:2007-01-20 Published:2007-01-20

Abstract: Nonlinear time-sequence analysis is used to study the time evolution of El Nino-Southern Oscillation (ENSO) by combining the method of global function approximation with lyapunov exponent analysis. The method includes phase-space reconstruction, lyapunov exponent analysis, global function approximation, principal components analysis and least-square estimation. The data used here is monthly sea surface temperature anomalies (SSTA) from CZ (Cane & Zebiak) model. It differs from traditional time-sequence analysis methods in which nonlinear chaotic time-sequence prediction is used. The present method is proved to be a successful one with lesser data, and it provides an alternative method for ENSO prediction.