Significant atmospheric nonlinearities in the ENSO cycle

The nonlinearities that cause El Nino events to deviate more from the mean state than La Nina events are still not completely understood. In this paper we investigate the contribution of one candidate mechanism: ENSO nonlinearities originating from the atmosphere. The initially linear Intermediate Complexity Model of the equatorial Pacific ocean in which all couplings were fitted to observations describes the ENSO cycle reasonably well. In this linear model we systematically introduce extra terms in the atmospheric component. These are the nonlinear response of mean wind stress to SST anomalies, the skewness of the driving noise term in the atmosphere and the relation of this noise term to the background SST or the ENSO phase. The nonlinear response of mean wind stress to SST in the ENSO region is found to be the dominant term influencing the ENSO cycle. However, this influence is only visible when noise fields are used that are fitted to observed patterns of prescribed standard deviation and spatial decorrelation. Standard deviation and skewness of noise do have a dependence on the ENSO phase, but this has a relatively small influence on the ENSO cycle in this model. With these additional nonlinearities in the representation of the atmosphere a step forward has been made towards building a realistic reduced complexity model for ENSO.