ENSO non-linearity in a changing climate

The El Niño Southern Oscillation (ENSO) phenomenon is characterized by a remarkable non-linearity (or asymmetry) concerning its spatial pattern: El Niño events are on average associated with stronger SST anomalies, more confined to the equator and located more in the east in comparison with La Niña events. State-of-the-art climate models still have problems in simulating the non-linearity of ENSO: only 14 of the 39 CMIP5 models simulate this spatial non-linearity quite realistically.

Here we address the issue of how the non-linearity may change in response to global warming. We analyze results from the CMIP5 data base and model experiments with the Kiel Climate Model (KCM) and address two major questions:

1. Are the projections of models, which in control runs realistically simulate the ENSO non-linearity, different to the projections of models that are not able to realistically simulate ENSO non-linearity?

2. What is important for a climate model to be capable of simulating ENSO non-linearity?

The projections of ENSO variability changes due to global warming in the RCP 8.5 scenario is quite different in the “linear” models compared to the “non-linear” models: the linear models project an increase in the Central Pacific (CP) El Niños and only a weak decrease in East Pacific (EP) El Niños. The non-linear models project no statistically significant change in CP El Niños and a strong decrease in EP El Niños.

We have additionally conducted some analysis on a set of experiments with the KCM, where we changed the horizontal and vertical resolution of the atmosphere model. In the standard configuration (T42L19), the KCM has a very linear ENSO. The strongest effect on the non-linearity has the increase in the vertical resolution. With a relative coarse horizontal resolution of T42 and 62 vertical levels, the KCM simulates the non-linearity of ENSO very well. The response of the wind circulation, especially the vertical wind, and convection to SST anomalies plays an important role, which both get more sensitive with higher vertical resolution. We conclude that the atmosphere plays an important role in the non-linearity of ENSO.