NUMERICAL SIMULATION OF THE ENSO CYCLE PART Ⅱ:──VARIABILITY AND TIME SCALE SELECTION MECHANISM

By using a tropical Pacific coupled air-sea model with an oceanic surface boundary layer and atmospheric internal convergence feedback effect,after a long time integration of 30 years,key features of the observed multiple time scale variability of ENSO cycle have been successfully simulated.Based on the simulated results a nonlinear analog model for understanding the formation of the principal period of ENSO cycle has been established and a possible time scale selection mechanism for ENSO cycle has also been proposed.It is reported that,1)similar to the observational facts,model ENSO cycle is truely a three time scale processes,i.e.the principal 3-4 year lower-frequency(LF) oscillation,quasi-biennial (QB) oscillation explaining the variance of 10%-200% and annual cycle (AC);2)the LF mode with 3-4 year principal period is a self-sustaining oscillation of nonlinear system.Its time scale selection results from a process of decreasing frequency and intensifying amplitude in the warm state of the QB mode,i.e.the fundamental intrinsic mode of the linear system,via the nonlinear mechanism,in which the single-way atmospheric convergence feedback appears to be mostly important to the process;3)although the annual cycle process can not essentially influence the formation of ENSO cycle,it can notably fix the phase and modulate the amplitude.Resultantly,the individual warm/cold events in ENSO cycle appear to be partly locked to the annual cycle,and also exhibit obvious irregularity both in amplitude and in lasting time.Therefore,ENSO variability or ENSO cycle is truely due to a nonlinear interaction of multiple time scales among the LF,QB and Ac modes,in which the QB mode is a fundamental mode.The time scale selection mechanism presented in this paper is more objective to approach real world ENSO process,because it can explain not only formation of the ENSO principal period but also the spectrum of ENSO variability.