Abstract
Abstract Decadal variations in the amplitude of El Niño and the Southern Oscillation have been the subject of great interest in the literature for the past decade. One theory suggests that ENSO is best described as a stable system driven by linear dynamics and that stochastic atmospheric forcing is responsible for the development and modulation of ENSO on interannual as well as decadal time scales. Another theory suggests that ENSO is driven by strong nonlinear coupled feedbacks between the ocean and atmosphere and low frequency changes in ENSO amplitude are driven by decadal changes in the tropical Pacific mean state. Unfortunately, the observed record is too short to collect reliable statistics for such low frequency behavior. A hybrid coupled model composed of a simple statistical atmosphere coupled to the Poseidon isopycnal ocean model has been developed for the study of ENSO decadal variability. The model simulates realistic ENSO variability on interannual and decadal time scales with negligible climate drift over 1000 years. Through analysis and experimentation the authors show that low frequency changes in the atmospheric “weather noise” drive changes in the tropical Pacific mean state leading to changes in the amplitude of ENSO on decadal time scales. Additional model simulations suggest that, while predictability is limited by the presence of atmospheric noise, there are extended periods when the coupled instability, strengthened by changes in the mean state, is insensitive to noise on interannual time scales. The relationship between decadal modulation of ENSO and mean state changes resides somewhere between the linear damped stochastically forced theory and the strongly unstable theory. Unlike the strongly unstable system, changes in ENSO amplitude on longer time scales are determined by the stochastic forcing. The stochastic forcing is not necessary in this model to sustain ENSO; however, its presence is crucial for low frequency changes in the mean state of the tropical Pacific. The strong relationship between the mean state and ENSO amplitude modulation in the model is in opposition to the linear damped stochastically forced theory. The fact that changes in the tropical Pacific mean state lead directly to changes in ENSO amplitude and predictability has positive implications for predictability.
