Nonlinearity and asymmetry in the atmospheric branch of ENSO with a simple GCM

<p>We assess dynamical nonlinearities and asymmetries in the atmospheric branch of the Bjerkness feedback. The atmospheric response to El Nino-related sea surface temperature anomalies (SSTAs), consists of developing deep convection with low-level westerlies west of the anomalous heating. Simple models of ENSO usually assume this process to be linear and symmetric. In this study, we use a global primitive equation model to evaluate nonlinearities and asymmetries that can potentially modify the Bjerkness mechanism.</p><p>The Dynamical Research Empirical Atmospheric Model (DREAM) is a global primitive equation model. It explicitly calculates advection and linear diffusion but physical processes are represented by an external empirical forcing deduced from ERA-Interim data. DREAM can be used as a stationary wave model where a perturbation develops on a fixed basic state. Both linear and nonlinear perturbations can be simulated by controlling the amplitude of the anomaly. DREAM can also be used as a simple GCM. In this case the model develops transient eddy fluxes and we obtain a full nonlinear solution. ENSO perturbations are introduced as a deep convective heating based on precipitation anomalies. Both idealised and realistic heating anomalies are studied, deduced from observed precipitation composites or calculated from SSTAs using an empirical transfer function.</p><p>The linear response to equatorial heating is strongly dependent on the position of the heating. Stationary equilibrium responses to idealised equatorial heating show that the response to a source in the East Pacific is larger than for the West Pacific. Stationary wave solutions are weakly nonlinear in heating amplitude, and the modification of the linear response is well explained as a linear response to stationary nonlinear fluxes. In fully turbulent simple GCM experiments, the response to equal and opposite heating perturbation shows a much larger asymmetry. This is diagnosed in terms of the linear response to anomalous transient eddy fluxes. The degree of nonlinearity in the response depends on the region and the variable under consideration. Nonlinearity is strong in the extratropics but less likely to be a significant factor locally and so probably of secondary importance for the Bjerkness feedback. However, the sensitivity of the response to diverse and asymmetric perturbation forcing implies that even when nonlinearity is weak, the asymmetric nature of the atmospheric response must be taken into account.</p>