A systematic approach to quantifying the noise-enso relationship and its effect on extreme el nino events

Abstract

State-dependent noise forcing, where properties such as amplitude are dependent on the background conditions, has been shown as a potential cause for the El Niño--La Niña amplitude asymmetry. Complex characterization of the state dependency of El Niño noise forcing as documented in most existing research has not been readily applicable to the outputs from the current generation of coupled climate models (GCMs). Here a simple method for determining the overall strength of the state-dependence factor of the El Niño noise forcing is proposed and tested. This method is shown to be independent of the data sampling from monthly mean to daily data, which make it readily applicable to monthly climate data archives. Using a reanalysis product and two coupled GCMs, the method is then applied to the equatorial zonal windstress, which is a known noise source for the El Niño-Southern Oscillation (ENSO) phenomenon. The windstress in all three cases is shown to have strong state-dependent noise forcing. The state-dependent component of the noise forcing is further isolated and shown to be enhanced by the Pacific Warm Pool and Westerly Wind Burst interaction. The coupled process acts to increase the low-frequency component of state-dependent windstress noise forcing, a part of noise forcing that is most essential in the excitation and strengthening of ENSO. The methodology is then further applied to the outputs from a suite of Coupled Model Intercomparison Project (CMIP)5 model simulations under various scenarios. It is found that most of the CMIP5 models underestimate the state-dependence factor of the equatorial windstress on sea-surface temperature (SST). There is a consistent relationship with the zonal equatorial SST gradient and the state-dependence factor. However, the models fail to agree on changes of the state-dependence factor due to global warming. Finally, using the conceptual model, the state dependence factor is shown to control the ENSO skewness and is a predictor of the frequency of occurrence of extreme El Niño events. To a large extent, the CMIP5 simulations are in support of these theoretical relationships.